*The problem:* AI search is exploding (Perplexity does 100M+ queries/month), but there's no way to know if these engines are citing your content. It's like SEO in 2005 before Google Analytics existed.
*What it does:*\n- Automated daily checks across multiple AI platforms\n- Tracks citation rate, position, and trends over time\n- AI-powered recommendations for queries where you're NOT cited\n- Shows which platforms cite you most
*Technical approach:*\n- Puppeteer for web scraping (no expensive APIs)\n- Firebase for real-time data\n- Groq for AI recommendations\n- Next.js + TypeScript frontend\n- Vercel cron for automation
*Why I built it:* I run llmscentral.com (a repository of llms.txt files) and noticed traffic from Perplexity was growing 50% month-over-month. But I had no idea which queries were driving it. Built this to scratch my own itch.
*Free tier:* 3 tracked queries, weekly checks on Perplexity\n*Paid tiers:* More queries, daily checks, all platforms, AI recommendations
Try it: https://llmscentral.com/citation-tracking
Would love feedback from the HN community! Happy to answer any technical questions."},"title":{"matchLevel":"none","matchedWords":[],"value":"Show HN: I built a tool to track when AI engines cite your website"}},"_tags":["story","author_legitcoders","story_45713722","show_hn"],"author":"legitcoders","created_at":"2025-10-26T17:37:57Z","created_at_i":1761500277,"num_comments":0,"objectID":"45713722","points":2,"story_id":45713722,"story_text":"Hey HN! I built a tool that tracks when AI search engines (Perplexity, ChatGPT, Claude, Google AI) cite your website in their responses.
*The problem:* AI search is exploding (Perplexity does 100M+ queries/month), but there's no way to know if these engines are citing your content. It's like SEO in 2005 before Google Analytics existed.
*What it does:*\n- Automated daily checks across multiple AI platforms\n- Tracks citation rate, position, and trends over time\n- AI-powered recommendations for queries where you're NOT cited\n- Shows which platforms cite you most
*Technical approach:*\n- Puppeteer for web scraping (no expensive APIs)\n- Firebase for real-time data\n- Groq for AI recommendations\n- Next.js + TypeScript frontend\n- Vercel cron for automation
*Why I built it:* I run llmscentral.com (a repository of llms.txt files) and noticed traffic from Perplexity was growing 50% month-over-month. But I had no idea which queries were driving it. Built this to scratch my own itch.
*Free tier:* 3 tracked queries, weekly checks on Perplexity\n*Paid tiers:* More queries, daily checks, all platforms, AI recommendations
Try it: https://llmscentral.com/citation-tracking
Would love feedback from the HN community! Happy to answer any technical questions.","title":"Show HN: I built a tool to track when AI engines cite your website","updated_at":"2026-03-05T22:55:27Z"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"chr15m"},"story_text":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["seo","data","api","expensive"],"value":"In April last year I started thinking about using Twitter in a smarter way. I wanted to do analytics on my tweets and find out more about people following me on Twitter. What kinds of things do people who follow me like and retweet? I decided to dig into the data and find out.
When I went to try and download Twitter data in raw form I found I quickly got bogged down writing API wrangling code and fiddling with API keys. I just wanted to crunch some data but here I was wrangling Twitter's API. This was such a frustrating experience it suddenly looked like an opportunity to me. Was there room for a product here? A product which does one simple thing: help people extract Twitter data painlessly without writing any code.
I did some research and discovered some tools that purported to do this, but none of them worked well for my use-case and all were badly designed and/or expensive. I decided to take a shot at it.
I worked on this as a side project in a "calm company" fashion. Each week for 26 weeks I would put aside one day to chip away at the features. I tried not to think about how the issue tracker was filling up more and more. Several times I pared back the feature set to try to really focus on the core use-case.
When my first user reached out and engaged I knew I might be onto something. I kept posting progress on Twitter and a few more people started to use it each week. Some of them came back, hinting at possible user retention. I hired a writer to write some articles to help with SEO and I kept working towards and MVP that I could use to test the market.
Finally the day arrived where all of the critical issues in my issue tracker were closed. That meant it was launch day. That was yesterday. So here I am releasing this on Hacker News to you, dear reader.
God speed little micro-SaaS, may the winds of fortune be at your back."},"title":{"fullyHighlighted":false,"matchLevel":"partial","matchedWords":["data","api"],"value":"Show HN: Download Twitter data without API keys"},"url":{"matchLevel":"none","matchedWords":[],"value":"https://tweetfeast.com/"}},"_tags":["story","author_chr15m","story_30188681","show_hn"],"author":"chr15m","children":[30204116,30208590,30208728,30208904,30209239,30209325,30209411,30209531,30209946,30212912,30216693,30219633],"created_at":"2022-02-03T04:28:21Z","created_at_i":1643862501,"num_comments":43,"objectID":"30188681","points":57,"story_id":30188681,"story_text":"In April last year I started thinking about using Twitter in a smarter way. I wanted to do analytics on my tweets and find out more about people following me on Twitter. What kinds of things do people who follow me like and retweet? I decided to dig into the data and find out.
When I went to try and download Twitter data in raw form I found I quickly got bogged down writing API wrangling code and fiddling with API keys. I just wanted to crunch some data but here I was wrangling Twitter's API. This was such a frustrating experience it suddenly looked like an opportunity to me. Was there room for a product here? A product which does one simple thing: help people extract Twitter data painlessly without writing any code.
I did some research and discovered some tools that purported to do this, but none of them worked well for my use-case and all were badly designed and/or expensive. I decided to take a shot at it.
I worked on this as a side project in a "calm company" fashion. Each week for 26 weeks I would put aside one day to chip away at the features. I tried not to think about how the issue tracker was filling up more and more. Several times I pared back the feature set to try to really focus on the core use-case.
When my first user reached out and engaged I knew I might be onto something. I kept posting progress on Twitter and a few more people started to use it each week. Some of them came back, hinting at possible user retention. I hired a writer to write some articles to help with SEO and I kept working towards and MVP that I could use to test the market.
Finally the day arrived where all of the critical issues in my issue tracker were closed. That meant it was launch day. That was yesterday. So here I am releasing this on Hacker News to you, dear reader.
God speed little micro-SaaS, may the winds of fortune be at your back.","title":"Show HN: Download Twitter data without API keys","updated_at":"2024-09-20T10:28:54Z","url":"https://tweetfeast.com/"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"poletopole"},"story_text":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["seo","data","api","expensive"],"value":"I was recently talking with my brother who is a SEO practitioner about making a way for him and others (whom aren't techies) to simply download a CSV of all his Google Search Console API data which is hard to do without a knowledge of programming. Personally, there's been many times in my career that I would pick up a backlogged bug item and have to dig around months prior for the data to fix it and wished that I could get permission from my boss to write a proper cache server to poll our API data but there was never a perceived time or place to do it. Often the bug couldn't be solved because the given API no longer had the data. Also it would be slow for certain APIs and tricky to make sure you don't go over your quota.
Mostly I think that developers such as myself would like an easy and affordable way to safely just save all their API data without having to write their own servers to transfer and cache and archive it and have it normalized with something like JSON-LD with a more sane API design, and perhaps use SQL or something like it to interface with their cached data. However, for non-technical users, I think a spreadsheet interface would be preferred and would open up your every day user to new world of possibilities. Either or, I feel what matters is taking ownership over your data (which can get expensive to retain with a given API) especially in these ransomware times. This sort of service I'm alluding to is easier to do with OAuth services than it is for say Twilio because this hypothetical service would need your API key which could be a deal breaker for some companies.
So if you could magically turn an API into a spreadsheet, what API(s) would those be and why?"},"title":{"fullyHighlighted":false,"matchLevel":"partial","matchedWords":["api"],"value":"Ask HN: What API do you wish were a spreadsheet?"}},"_tags":["story","author_poletopole","story_27238170","ask_hn"],"author":"poletopole","children":[27242988],"created_at":"2021-05-21T18:44:05Z","created_at_i":1621622645,"num_comments":1,"objectID":"27238170","points":2,"story_id":27238170,"story_text":"I was recently talking with my brother who is a SEO practitioner about making a way for him and others (whom aren't techies) to simply download a CSV of all his Google Search Console API data which is hard to do without a knowledge of programming. Personally, there's been many times in my career that I would pick up a backlogged bug item and have to dig around months prior for the data to fix it and wished that I could get permission from my boss to write a proper cache server to poll our API data but there was never a perceived time or place to do it. Often the bug couldn't be solved because the given API no longer had the data. Also it would be slow for certain APIs and tricky to make sure you don't go over your quota.
Mostly I think that developers such as myself would like an easy and affordable way to safely just save all their API data without having to write their own servers to transfer and cache and archive it and have it normalized with something like JSON-LD with a more sane API design, and perhaps use SQL or something like it to interface with their cached data. However, for non-technical users, I think a spreadsheet interface would be preferred and would open up your every day user to new world of possibilities. Either or, I feel what matters is taking ownership over your data (which can get expensive to retain with a given API) especially in these ransomware times. This sort of service I'm alluding to is easier to do with OAuth services than it is for say Twilio because this hypothetical service would need your API key which could be a deal breaker for some companies.
So if you could magically turn an API into a spreadsheet, what API(s) would those be and why?","title":"Ask HN: What API do you wish were a spreadsheet?","updated_at":"2024-09-20T08:36:32Z"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"solardev"},"comment_text":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["seo","data","api","expensive"],"value":"> They already have a WordPress/Woocommerce solution that is slow and very inflexible for what they need.
I concur with the other poster that it's probably easier/safer to fix and/or extend the Woocommerce with additional functionality than to try to reinvent the wheel from scratch. Wordpress/PHP is pretty extensible and can be made performant with good caching.
Probably moving it to a better Wordpress host like Pantheon or WPEngine would speed it up with minimal effort (but possibly high hosting costs, which are usually still cheaper than dev time).
Depending on what additional flexibility you need, you might be able to write plugins to add just that functionality while keeping the basic ecommerce stuff in the core?
> Good thing is I will not be hiring any developers, the plan is to hire a development firm that will build out the website.
It would probably help if the agency has prior ecommerce experience with a popular stack. It doesn't have to be Woocommerce or Shopify specifically, but I'd stay away from something totally custom unless you absolutely have to. A commodity stack is a lot easier to maintain over time and to find people to work on later if they need to add additional features. Even something no-code/low-code like Wix/SquareSpace/Square these days has ecommerce built-in, and that's a heck lot less work than building a web app from scratch with no experience. At the very least, you should outsource payment processing at least, like to Stripe or others, instead of processing your own credit card payments using custom software (PCI compliance is hard and it's easy to accidentally create vulnerabilities)... you should never know your customers' credit card numbers if you can help it.
If I were doing this, I'd really look at the business (profit margins, how they handle orders and shipping and refunds and all that), who's on the dev team both now and for the foreseeable future, etc.
It basically comes down to how much staff the business wants to keep over time to maintain this thing in-house, vs outsourcing as much of possible to a specialized vendor that they can just have a contractual relationship with.
From one techie to another, this is the sort of thing any of us can build with some time and research, but that's the sort of project that can too easily backfire and end up creating a lot of tech debt for your client (your friend) if nobody else can easily maintain it later.
If your client uses a run of the mill Woocommerce setup, after you leave they can easily hire any ol' Wordpress dev for a few days to fix whatever comes up. That might cost a few hundred dollars.
But if you make something completely custom for them using a nonstandard stack, suddenly it's a multi-month affair for a much more experienced (and expensive) dev to come in and reverse engineer your whole stack and learn to extend it, only to extend it with their own nonstandard way. Suddenly this simple ecommerce site is now a full-blown web app requiring a whole team of specialists to maintain. Now it's several tens of thousands of dollars. It's not good for most small businesses to have to deal with something like that.
IMHO a commodity online shop shouldn't be a fancy vanity project; it should be simple and reliable and predictable and require the least amount of ongoing maintenance possible.
> Thanks for the info on the intricacies, do you have any resources that can help me out with those?
Sorry, I'm not the best person to ask that of :( I've mostly just learned through accumulated experience working on ecommerce and other sites.
I bet these days there are good online tutorials or YouTube videos to get you started, but an ecommerce shop is such a broad (if relatively shallow) app that it's hard to get it all right. It really depends on the level of functionality, security, etc. that you need, and at what budget, with what staff/expertise availabile.
At the simplest level, if you can outsource all actual order management and fulfillment to a third party (like Amazon or Shopify or whatever), you basically end up just creating a bunch of mostly static product pages that pulls pricing and order data from an API. That's not so bad since all the critical stuff is handled elsewhere.
But the moment you need to start maintaining your own databases for all that stuff, it can get a lot more complicated really quickly.
There are also partially-vendored solutions like CommerceLayer or Commerce.js that do some of the heavy lifting for you to add a simple frontend on top of, or old-school PHP monoliths like Magento (now owned by Adobe, apparently).
I don't know if you can afford this, but it might help to look at some high-level (abstract) ecommerce UX, like from the UX group NN: https://www.nngroup.com/reports/ecommerce-user-experience/ This discusses many of the user-facing concerns that you can think about whether you need to build. If you don't want to buy the report, you can also browse through their free articles and videos: https://www.nngroup.com/topic/e-commerce/. That just discusses the UX though and not so much the "how to build it" part.
If you've never done web dev before, some random keywords you can look into for an idea of what you might have to deal with are: payment gateways/hosted checkouts, iframes, CORS, TLS/HTTPS, PCI, AJAX, ACID, accessibility, internationalization, CDNs, responsiveness, dev/staging envs (or CI/CD for something more enterprise-y), network stuff (DNS, hosting, caching, load balancers if you get a lot of traffic), SEO, forms & validations & sanitization, roles & permissions & session management, LEMP vs Jamstack, CRM, transactional emails, SPF/DKIM/DMARC (email deliverability)... that's probably a good start? None of those topics are individually very difficult, it's just that there's so many of them to juggle together to make a basic, functional ecommerce solution :("},"story_title":{"matchLevel":"none","matchedWords":[],"value":"Ask HN: How can I oversee an e-com website development?"}},"_tags":["comment","author_solardev","story_41063650"],"author":"solardev","children":[41074618],"comment_text":"> They already have a WordPress/Woocommerce solution that is slow and very inflexible for what they need.
I concur with the other poster that it's probably easier/safer to fix and/or extend the Woocommerce with additional functionality than to try to reinvent the wheel from scratch. Wordpress/PHP is pretty extensible and can be made performant with good caching.
Probably moving it to a better Wordpress host like Pantheon or WPEngine would speed it up with minimal effort (but possibly high hosting costs, which are usually still cheaper than dev time).
Depending on what additional flexibility you need, you might be able to write plugins to add just that functionality while keeping the basic ecommerce stuff in the core?
> Good thing is I will not be hiring any developers, the plan is to hire a development firm that will build out the website.
It would probably help if the agency has prior ecommerce experience with a popular stack. It doesn't have to be Woocommerce or Shopify specifically, but I'd stay away from something totally custom unless you absolutely have to. A commodity stack is a lot easier to maintain over time and to find people to work on later if they need to add additional features. Even something no-code/low-code like Wix/SquareSpace/Square these days has ecommerce built-in, and that's a heck lot less work than building a web app from scratch with no experience. At the very least, you should outsource payment processing at least, like to Stripe or others, instead of processing your own credit card payments using custom software (PCI compliance is hard and it's easy to accidentally create vulnerabilities)... you should never know your customers' credit card numbers if you can help it.
If I were doing this, I'd really look at the business (profit margins, how they handle orders and shipping and refunds and all that), who's on the dev team both now and for the foreseeable future, etc.
It basically comes down to how much staff the business wants to keep over time to maintain this thing in-house, vs outsourcing as much of possible to a specialized vendor that they can just have a contractual relationship with.
From one techie to another, this is the sort of thing any of us can build with some time and research, but that's the sort of project that can too easily backfire and end up creating a lot of tech debt for your client (your friend) if nobody else can easily maintain it later.
If your client uses a run of the mill Woocommerce setup, after you leave they can easily hire any ol' Wordpress dev for a few days to fix whatever comes up. That might cost a few hundred dollars.
But if you make something completely custom for them using a nonstandard stack, suddenly it's a multi-month affair for a much more experienced (and expensive) dev to come in and reverse engineer your whole stack and learn to extend it, only to extend it with their own nonstandard way. Suddenly this simple ecommerce site is now a full-blown web app requiring a whole team of specialists to maintain. Now it's several tens of thousands of dollars. It's not good for most small businesses to have to deal with something like that.
IMHO a commodity online shop shouldn't be a fancy vanity project; it should be simple and reliable and predictable and require the least amount of ongoing maintenance possible.
> Thanks for the info on the intricacies, do you have any resources that can help me out with those?
Sorry, I'm not the best person to ask that of :( I've mostly just learned through accumulated experience working on ecommerce and other sites.
I bet these days there are good online tutorials or YouTube videos to get you started, but an ecommerce shop is such a broad (if relatively shallow) app that it's hard to get it all right. It really depends on the level of functionality, security, etc. that you need, and at what budget, with what staff/expertise availabile.
At the simplest level, if you can outsource all actual order management and fulfillment to a third party (like Amazon or Shopify or whatever), you basically end up just creating a bunch of mostly static product pages that pulls pricing and order data from an API. That's not so bad since all the critical stuff is handled elsewhere.
But the moment you need to start maintaining your own databases for all that stuff, it can get a lot more complicated really quickly.
There are also partially-vendored solutions like CommerceLayer or Commerce.js that do some of the heavy lifting for you to add a simple frontend on top of, or old-school PHP monoliths like Magento (now owned by Adobe, apparently).
I don't know if you can afford this, but it might help to look at some high-level (abstract) ecommerce UX, like from the UX group NN: https://www.nngroup.com/reports/ecommerce-user-experience/ This discusses many of the user-facing concerns that you can think about whether you need to build. If you don't want to buy the report, you can also browse through their free articles and videos: https://www.nngroup.com/topic/e-commerce/. That just discusses the UX though and not so much the "how to build it" part.
If you've never done web dev before, some random keywords you can look into for an idea of what you might have to deal with are: payment gateways/hosted checkouts, iframes, CORS, TLS/HTTPS, PCI, AJAX, ACID, accessibility, internationalization, CDNs, responsiveness, dev/staging envs (or CI/CD for something more enterprise-y), network stuff (DNS, hosting, caching, load balancers if you get a lot of traffic), SEO, forms & validations & sanitization, roles & permissions & session management, LEMP vs Jamstack, CRM, transactional emails, SPF/DKIM/DMARC (email deliverability)... that's probably a good start? None of those topics are individually very difficult, it's just that there's so many of them to juggle together to make a basic, functional ecommerce solution :(","created_at":"2024-07-25T22:46:08Z","created_at_i":1721947568,"objectID":"41074256","parent_id":41070173,"story_id":41063650,"story_title":"Ask HN: How can I oversee an e-com website development?","updated_at":"2025-07-18T20:47:56Z"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"jppope"},"comment_text":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["seo","data","api","expensive"],"value":"yes, LLMs will have a massive impact on SEO
but the dirty secret here... GPT is trained pretty heavily on reddit data. theres a reason why there hasn't been an IPO and their api got crazy expensive. OpenAi is trying to build a moat with Reddit and with regulation. I really don't think its going to be that long till people figure it out, but who knows"},"story_title":{"fullyHighlighted":false,"matchLevel":"partial","matchedWords":["seo"],"value":"As the public begins to believe Google isn\u2019t as useful, what happens to SEO?"},"story_url":{"fullyHighlighted":false,"matchLevel":"partial","matchedWords":["seo"],"value":"https://www.theverge.com/features/23931789/seo-search-engine-optimization-experts-google-results"}},"_tags":["comment","author_jppope","story_38097938"],"author":"jppope","comment_text":"yes, LLMs will have a massive impact on SEO
but the dirty secret here... GPT is trained pretty heavily on reddit data. theres a reason why there hasn't been an IPO and their api got crazy expensive. OpenAi is trying to build a moat with Reddit and with regulation. I really don't think its going to be that long till people figure it out, but who knows","created_at":"2023-11-02T05:06:15Z","created_at_i":1698901575,"objectID":"38109228","parent_id":38097938,"story_id":38097938,"story_title":"As the public begins to believe Google isn\u2019t as useful, what happens to SEO?","story_url":"https://www.theverge.com/features/23931789/seo-search-engine-optimization-experts-google-results","updated_at":"2024-09-20T15:32:27Z"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"alex_bob"},"story_text":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["seo","data","api","expensive"],"value":"Hi HN \u2013 I\u2019m a marketer with 10+ years of experience. I used to work in big companies, with big brands and big budgets. We had full creative teams, external agencies, media buyers, and enough money to test everything.
Back then, I spent $10\u201320k a day on ads. If something didn\u2019t work, we just pivoted fast.
Three years ago, I switched to early-stage B2B SaaS. Suddenly, I had to do everything \u2014 with no money, no team, and no margin for error.
And I hated it.
The hardest part? Getting attention.
CPCs are now $5\u201315 even in niche B2B. SEO doesn\u2019t perform like it used to (AI spam everywhere). Paid channels are expensive and feel like a black hole.
So I turned to LinkedIn.
It\u2019s weird, but still effective. Everyone in B2B is there \u2014 decision-makers, investors, customers. I saw other founders growing personal brands and getting traction. I tried it too.
But it felt awkward. Posts took too long to write. I had no content ideas. Most posts sounded cringe. I skipped posting all the time.
So I built a tool for myself.
Now it\u2019s a product: Viralbuddy.io
It\u2019s for early-stage founders and marketers who want to grow on LinkedIn without wasting hours.
It helps with:
Ideas: I scraped a library of viral LinkedIn posts and structured them. You can use them as-is or regenerate them with AI for your audience.
Writing: I spent a lot of time refining prompts using OpenAI and Claude. Posts are tailored to your goal (clients, investors, team, etc.) and follow a solid hook\u2013body\u2013CTA format.
Content creation: I built image and carousel generation using Recraft to save time and avoid design bottlenecks.
Now I post 3\u20135 times a week. My network grows by 300\u2013500 people monthly. I\u2019ve gotten beta users, partner intros, and even investor meetings \u2014 all through personal posts.
It\u2019s not a magic bullet. You still need to show up and write honestly. But it helps.
How it\u2019s built:
Scraping viral post data: Apify
Content generation: OpenAI (ChatGPT API) + Claude (Anthropic)
Image & carousel generation: Recraft
Frontend: basic onboarding logic + prompt flow in JS
Landing page: WordPress, set up with a $50 template + $50/year hosting
I\u2019d love feedback, questions, or criticism. Happy to share how I built the prompt logic, YouTube summarizer, or the post structuring method.
Thanks!"},"title":{"matchLevel":"none","matchedWords":[],"value":"Show HN: I built a tool to make early-stage marketing on LinkedIn less painful"},"url":{"matchLevel":"none","matchedWords":[],"value":"https://www.viralbuddy.io/"}},"_tags":["story","author_alex_bob","story_44145093","show_hn"],"author":"alex_bob","created_at":"2025-05-31T15:56:03Z","created_at_i":1748706963,"num_comments":0,"objectID":"44145093","points":1,"story_id":44145093,"story_text":"Hi HN \u2013 I\u2019m a marketer with 10+ years of experience. I used to work in big companies, with big brands and big budgets. We had full creative teams, external agencies, media buyers, and enough money to test everything.
Back then, I spent $10\u201320k a day on ads. If something didn\u2019t work, we just pivoted fast.
Three years ago, I switched to early-stage B2B SaaS. Suddenly, I had to do everything \u2014 with no money, no team, and no margin for error.
And I hated it.
The hardest part? Getting attention.
CPCs are now $5\u201315 even in niche B2B. SEO doesn\u2019t perform like it used to (AI spam everywhere). Paid channels are expensive and feel like a black hole.
So I turned to LinkedIn.
It\u2019s weird, but still effective. Everyone in B2B is there \u2014 decision-makers, investors, customers. I saw other founders growing personal brands and getting traction. I tried it too.
But it felt awkward. Posts took too long to write. I had no content ideas. Most posts sounded cringe. I skipped posting all the time.
So I built a tool for myself.
Now it\u2019s a product: Viralbuddy.io
It\u2019s for early-stage founders and marketers who want to grow on LinkedIn without wasting hours.
It helps with:
Ideas: I scraped a library of viral LinkedIn posts and structured them. You can use them as-is or regenerate them with AI for your audience.
Writing: I spent a lot of time refining prompts using OpenAI and Claude. Posts are tailored to your goal (clients, investors, team, etc.) and follow a solid hook\u2013body\u2013CTA format.
Content creation: I built image and carousel generation using Recraft to save time and avoid design bottlenecks.
Now I post 3\u20135 times a week. My network grows by 300\u2013500 people monthly. I\u2019ve gotten beta users, partner intros, and even investor meetings \u2014 all through personal posts.
It\u2019s not a magic bullet. You still need to show up and write honestly. But it helps.
How it\u2019s built:
Scraping viral post data: Apify
Content generation: OpenAI (ChatGPT API) + Claude (Anthropic)
Image & carousel generation: Recraft
Frontend: basic onboarding logic + prompt flow in JS
Landing page: WordPress, set up with a $50 template + $50/year hosting
I\u2019d love feedback, questions, or criticism. Happy to share how I built the prompt logic, YouTube summarizer, or the post structuring method.
Thanks!","title":"Show HN: I built a tool to make early-stage marketing on LinkedIn less painful","updated_at":"2025-05-31T15:58:04Z","url":"https://www.viralbuddy.io/"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"xohails"},"story_text":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["seo","data","api","expensive"],"value":"After spending years paying for dozens of SaaS tools (and watching prices constantly increase), I built an alternative that runs entirely in Google Sheets.
It started as a personal project to automate my own business tasks, but it's now helping thousands of businesses ditch expensive subscriptions.
The core idea is simple: Instead of paying monthly for scattered tools, why not leverage the power of Google Sheets as an automation hub?
What you can do without leaving Google Sheets:
\u2022 Build complete WordPress sites (no coding)\n\u2022 Generate AI content at scale\n\u2022 Automate SEO workflows\n\u2022 Manage Shopify/WooCommerce stores\n\u2022 Run email campaigns\n\u2022 Scrape and analyze web data
Technical Implementation:\n\u2022 Built on Google Apps Script for seamless integration\n\u2022 Direct API connections to WordPress, Shopify, OpenAI\n\u2022 Custom middleware for handling rate limits and data processing\n\u2022 Zero external dependencies - everything runs in your sheet
Real Numbers:
\u2022 Replaces 15+ separate SaaS tools\n\u2022 Average setup time: 15 minutes\n\u2022 Typical savings: $3,000-$5,000/year\n\u2022 One-time payment vs. monthly subscriptions\n\u2022 5,000+ businesses automated
The Secret Sauce:\nInstead of building yet another platform, we turned Google Sheets into one. Users keep working in a familiar environment, maintain data ownership, and use their own API keys. No learning curve, no lock-in.
Open Questions for HN:
What other APIs should we integrate? (Currently supporting OpenAI, WordPress, Shopify, WooCommerce)\nThoughts on using Google Sheets as a business automation platform?\nSecurity considerations when handling sensitive data in Sheets?
Try it: https://smartspreadsheets.co
Currently on AppSumo (lifetime deal): https://appsumo.com/products/smart-spreadsheets/
I'm the founder and will be around to answer questions and discuss the technical details."},"title":{"matchLevel":"none","matchedWords":[],"value":"Show HN: I Built a Google Sheets Alternative to $50K/Year in SaaS Subscriptions"},"url":{"matchLevel":"none","matchedWords":[],"value":"https://smartspreadsheets.co/"}},"_tags":["story","author_xohails","story_43555364","show_hn"],"author":"xohails","created_at":"2025-04-02T10:40:12Z","created_at_i":1743590412,"num_comments":0,"objectID":"43555364","points":1,"story_id":43555364,"story_text":"After spending years paying for dozens of SaaS tools (and watching prices constantly increase), I built an alternative that runs entirely in Google Sheets.
It started as a personal project to automate my own business tasks, but it's now helping thousands of businesses ditch expensive subscriptions.
The core idea is simple: Instead of paying monthly for scattered tools, why not leverage the power of Google Sheets as an automation hub?
What you can do without leaving Google Sheets:
\u2022 Build complete WordPress sites (no coding)\n\u2022 Generate AI content at scale\n\u2022 Automate SEO workflows\n\u2022 Manage Shopify/WooCommerce stores\n\u2022 Run email campaigns\n\u2022 Scrape and analyze web data
Technical Implementation:\n\u2022 Built on Google Apps Script for seamless integration\n\u2022 Direct API connections to WordPress, Shopify, OpenAI\n\u2022 Custom middleware for handling rate limits and data processing\n\u2022 Zero external dependencies - everything runs in your sheet
Real Numbers:
\u2022 Replaces 15+ separate SaaS tools\n\u2022 Average setup time: 15 minutes\n\u2022 Typical savings: $3,000-$5,000/year\n\u2022 One-time payment vs. monthly subscriptions\n\u2022 5,000+ businesses automated
The Secret Sauce:\nInstead of building yet another platform, we turned Google Sheets into one. Users keep working in a familiar environment, maintain data ownership, and use their own API keys. No learning curve, no lock-in.
Open Questions for HN:
What other APIs should we integrate? (Currently supporting OpenAI, WordPress, Shopify, WooCommerce)\nThoughts on using Google Sheets as a business automation platform?\nSecurity considerations when handling sensitive data in Sheets?
Try it: https://smartspreadsheets.co
Currently on AppSumo (lifetime deal): https://appsumo.com/products/smart-spreadsheets/
I'm the founder and will be around to answer questions and discuss the technical details.","title":"Show HN: I Built a Google Sheets Alternative to $50K/Year in SaaS Subscriptions","updated_at":"2025-04-02T10:42:54Z","url":"https://smartspreadsheets.co/"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"Aurornis"},"comment_text":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["seo","data","api","expensive"],"value":"> So as long as they're winning, every corp is going to either shut down APIs or absurdly gouge prices. This is the new internet that we've voted for with money and attention.
I think you\u2019re missing the real cause of this shift: These free APIs existed during the investment-fueled growth phase, then disappeared when they started switching into the real business mode.
We had an unusually long period of time where companies could play the startup game of spending money and headcount on things that didn\u2019t generate much or any revenue. Free APIs were an artifact of that. The disappearance of the Twitter and Reddit APIs coincides with them shifting toward profitability.
You don\u2019t have to \u201cvote for\u201d anything for this to happen. When it stops being easy to run companies at a loss year over year, the parts of the company that aren\u2019t generating more revenue than they cost either get price increases or dropped entirely.
I\u2019ve been at a startup-like company going through this change. It was sad, though not at all surprising, when management started taking inventory of everything people were working on and cross referencing it with how much that was generating in revenue. There were a few moments of internal revelation when someone realized that entire teams of expensive engineers had been working on features and products that either very few people used or that were very complex but generated no revenue. It doesn\u2019t make business sense.
The abuse landscape has also changed dramatically. In the early days, free APIs were rarely used features by a few power users. Now, any free API is guaranteed to be abused by some growth hacking startup who wants to vacuum up all of your data and use it for SEO spam, AI training, or other purposes."},"story_title":{"fullyHighlighted":false,"matchLevel":"partial","matchedWords":["api"],"value":"Spotify has shut down several API endpoints"},"story_url":{"fullyHighlighted":false,"matchLevel":"partial","matchedWords":["api"],"value":"https://developer.spotify.com/blog/2024-11-27-changes-to-the-web-api"}},"_tags":["comment","author_Aurornis","story_42260481"],"author":"Aurornis","children":[42274697,42275266,42312813],"comment_text":"> So as long as they're winning, every corp is going to either shut down APIs or absurdly gouge prices. This is the new internet that we've voted for with money and attention.
I think you\u2019re missing the real cause of this shift: These free APIs existed during the investment-fueled growth phase, then disappeared when they started switching into the real business mode.
We had an unusually long period of time where companies could play the startup game of spending money and headcount on things that didn\u2019t generate much or any revenue. Free APIs were an artifact of that. The disappearance of the Twitter and Reddit APIs coincides with them shifting toward profitability.
You don\u2019t have to \u201cvote for\u201d anything for this to happen. When it stops being easy to run companies at a loss year over year, the parts of the company that aren\u2019t generating more revenue than they cost either get price increases or dropped entirely.
I\u2019ve been at a startup-like company going through this change. It was sad, though not at all surprising, when management started taking inventory of everything people were working on and cross referencing it with how much that was generating in revenue. There were a few moments of internal revelation when someone realized that entire teams of expensive engineers had been working on features and products that either very few people used or that were very complex but generated no revenue. It doesn\u2019t make business sense.
The abuse landscape has also changed dramatically. In the early days, free APIs were rarely used features by a few power users. Now, any free API is guaranteed to be abused by some growth hacking startup who wants to vacuum up all of your data and use it for SEO spam, AI training, or other purposes.","created_at":"2024-11-29T15:41:18Z","created_at_i":1732894878,"objectID":"42274581","parent_id":42274116,"story_id":42260481,"story_title":"Spotify has shut down several API endpoints","story_url":"https://developer.spotify.com/blog/2024-11-27-changes-to-the-web-api","updated_at":"2024-12-11T14:48:41Z"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"hiruben"},"story_text":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["seo","data","api","expensive"],"value":"As someone who's been in the content marketing space for years, I've watched the explosion of AI writing tools with mixed feelings. Sure, ChatGPT and similar tools can generate text quickly, but there's a fundamental problem: they produce content that looks AI-generated and doesn't rank well on Google.
Most AI writing tools are glorified text generators. They don't understand SEO, they don't help with keyword research, and they certainly don't know how to build the E-E-A-T (Experience, Expertise, Authoritativeness, Trustworthiness) signals that Google actually cares about.
What Makes Writer J Different\nWriter J isn't just another AI writing tool. It's a complete content creation workflow designed specifically for SEO and content marketing professionals. Here's what sets it apart:
1. Keyword Research Integration\nInstead of guessing what to write about, Writer J analyzes SERPs, identifies keyword gaps, and suggests content angles based on actual search data. No more shooting in the dark.
2. Credible Source Weaving\nThe tool automatically finds and integrates authoritative sources into your content. This isn't just citation \u2013 it's intelligent content synthesis that makes your articles more credible and link-worthy.
3. Reddit Insights Mining\nThis is probably my favorite feature. Writer J scans Reddit discussions to find real user pain points and questions. Your content addresses actual problems people are discussing, not what you think they want to read.
4. Natural Product Integration\nIf you're doing content marketing for a product (which most of us are), Writer J knows how to mention your product naturally without sounding like a sales pitch. It's subtle but effective.
5. E-E-A-T Authority Building\nGoogle's algorithm heavily weights expertise and authoritativeness. Writer J helps you build these signals by incorporating author credentials, experience markers, and trust indicators throughout your content.
6. Publish-Ready Output\nHere's the kicker \u2013 the content comes out ready to publish. No extensive editing needed. Proper formatting, SEO optimization, logical flow. Copy, paste, publish.
The Technical Implementation\nFor the developers in the audience, here's how we built this:
Multi-AI Architecture: We use Google Gemini as the primary model with DeepSeek as backup, allowing us to optimize for both quality and cost.
SERP Analysis: Real-time search engine results analysis using the Serper API to understand competitive landscape.
Batch Processing: The system can handle bulk content generation with intelligent queuing and rate limiting.
Why I Built This\nI was frustrated with the content treadmill. Clients wanted more content, faster, but Google kept raising the bar for quality. The existing AI tools felt like expensive autocomplete \u2013 they could write, but they couldn't think strategically about content.
After analyzing what actually ranks on Google (spoiler: it's not just well-written content), I realized we needed a tool that understood the entire content ecosystem \u2013 keywords, competition, user intent, authority signals, the works.
Looking for Feedback\nWe're launching publicly and looking for early users to stress-test the system and provide feedback. The tool is currently free while we gather user insights and improve the product.
What I'm particularly interested in feedback on:
Content Quality: How does the output compare to human-written content?\nWorkflow Efficiency: Does the 7-step process feel natural or cumbersome?\nSEO Results: Are you seeing actual ranking improvements?\nFeature Gaps: What's missing that would make this indispensable for your workflow?\nThe entire system is designed to scale \u2013 we're processing hundreds of articles daily without breaking a sweat.
TL;DR: Built an AI content generator that actually understands SEO and produces content that ranks. Looking for feedback from the HN community.
website: https://writer-j.com"},"title":{"fullyHighlighted":false,"matchLevel":"partial","matchedWords":["seo"],"value":"Show HN: Writer J \u2013 AI Content Generator with 7-Step SEO Workflow"},"url":{"matchLevel":"none","matchedWords":[],"value":"https://www.writer-j.com/"}},"_tags":["story","author_hiruben","story_44365036","show_hn"],"author":"hiruben","created_at":"2025-06-24T11:35:39Z","created_at_i":1750764939,"num_comments":0,"objectID":"44365036","points":1,"story_id":44365036,"story_text":"As someone who's been in the content marketing space for years, I've watched the explosion of AI writing tools with mixed feelings. Sure, ChatGPT and similar tools can generate text quickly, but there's a fundamental problem: they produce content that looks AI-generated and doesn't rank well on Google.
Most AI writing tools are glorified text generators. They don't understand SEO, they don't help with keyword research, and they certainly don't know how to build the E-E-A-T (Experience, Expertise, Authoritativeness, Trustworthiness) signals that Google actually cares about.
What Makes Writer J Different\nWriter J isn't just another AI writing tool. It's a complete content creation workflow designed specifically for SEO and content marketing professionals. Here's what sets it apart:
1. Keyword Research Integration\nInstead of guessing what to write about, Writer J analyzes SERPs, identifies keyword gaps, and suggests content angles based on actual search data. No more shooting in the dark.
2. Credible Source Weaving\nThe tool automatically finds and integrates authoritative sources into your content. This isn't just citation \u2013 it's intelligent content synthesis that makes your articles more credible and link-worthy.
3. Reddit Insights Mining\nThis is probably my favorite feature. Writer J scans Reddit discussions to find real user pain points and questions. Your content addresses actual problems people are discussing, not what you think they want to read.
4. Natural Product Integration\nIf you're doing content marketing for a product (which most of us are), Writer J knows how to mention your product naturally without sounding like a sales pitch. It's subtle but effective.
5. E-E-A-T Authority Building\nGoogle's algorithm heavily weights expertise and authoritativeness. Writer J helps you build these signals by incorporating author credentials, experience markers, and trust indicators throughout your content.
6. Publish-Ready Output\nHere's the kicker \u2013 the content comes out ready to publish. No extensive editing needed. Proper formatting, SEO optimization, logical flow. Copy, paste, publish.
The Technical Implementation\nFor the developers in the audience, here's how we built this:
Multi-AI Architecture: We use Google Gemini as the primary model with DeepSeek as backup, allowing us to optimize for both quality and cost.
SERP Analysis: Real-time search engine results analysis using the Serper API to understand competitive landscape.
Batch Processing: The system can handle bulk content generation with intelligent queuing and rate limiting.
Why I Built This\nI was frustrated with the content treadmill. Clients wanted more content, faster, but Google kept raising the bar for quality. The existing AI tools felt like expensive autocomplete \u2013 they could write, but they couldn't think strategically about content.
After analyzing what actually ranks on Google (spoiler: it's not just well-written content), I realized we needed a tool that understood the entire content ecosystem \u2013 keywords, competition, user intent, authority signals, the works.
Looking for Feedback\nWe're launching publicly and looking for early users to stress-test the system and provide feedback. The tool is currently free while we gather user insights and improve the product.
What I'm particularly interested in feedback on:
Content Quality: How does the output compare to human-written content?\nWorkflow Efficiency: Does the 7-step process feel natural or cumbersome?\nSEO Results: Are you seeing actual ranking improvements?\nFeature Gaps: What's missing that would make this indispensable for your workflow?\nThe entire system is designed to scale \u2013 we're processing hundreds of articles daily without breaking a sweat.
TL;DR: Built an AI content generator that actually understands SEO and produces content that ranks. Looking for feedback from the HN community.
website: https://writer-j.com","title":"Show HN: Writer J \u2013 AI Content Generator with 7-Step SEO Workflow","updated_at":"2025-06-24T11:55:07Z","url":"https://www.writer-j.com/"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"throwaway_burn"},"story_text":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["seo","data","api","expensive"],"value":"It's a commonly heard phrase in my office.
I'm the only developer where I work, based in a [car] dealership, which is one of the franchise locations of a car dealer group which owns [more than a few] (or so) other dealerships in quite a large area of the United Kingdom. For the record, I am not writing this on company time [half an hour lunch break, although he still does this even if I AM on my lunch break]
I build websites, web apps, microsites, soon to be Android Apps and iPhone apps. I am in charge of slicing and dicing photoshop templates which we get from our designer (Who sits behind me in the same office) into HTML templates to use with Drupal, CakePHP et al. I am in charge of inputting the copy for said microsites and websites from MS Word documents, this copy is written by the \"Marketing Manager\", who currently sits behind and to the right of me. To the right of where I sit, is my boss, the son of one of the Managing Directors of this company.
Do not get the tone of this post wrong, I love my job. I love the people, I love the fact that it's only twenty minutes down the road from my house, I love that parking is provided (although, not on site, about a quarter of a mile down the road). I even love my boss, he's a good, capable man and just wants to get things done and do what's best for this company.
However, as my work load has increased over the last two years, and things have gotten more difficult for me (harder problems, taking longer to solve and a lot more thinking) my brain has gotten tired. I've become burnt out maybe three times in the last two years, and have had to take a week off in some cases just to recover from the last project. (Considering we get 20 days holiday and zero sick pay, whenever I AM actually sick, I need to take holiday or I don't get paid for those days).
Sometimes, when I have really lost focus, or a problem has just circled in my head for too long to be solved without taking a break from thinking about it, I'll browse HN or reddit, not for very long, just for about 5 or 10 minutes until my head clears. More often than not, I'll be searching for the problem I have and reading articles about it.
My boss has the tremendous habit (when he is actually in the dealership), of coming into the office and looking straight at my screen, with the now dreaded words, \"Is that work?\", usually followed by \"Come on [throwaway_burn]! Focus! Get some fucking work done!\".
Now, my boss is a very visual person. If he can't instantly see results, he assumed I'm not doing any work. I've just spend four or five weeks creating a completely bespoke front end for building car quotes based on a completely experimental Web Service API built by a car valuation service. They usually make you host an MSSQL server, and they push or replicate their data to your server so you always have an up to date set of data to build cars with. My boss decided that this wasn't the way to go, because it's quite expensive so usually a single dealer group couldn't afford it, hence why you have aggregation systems, and people have it built into their customer management systems. [The web service company] probably lets us use this system for cheap as it is so undocumented when I first started using it, and it very slow, which means I have to pull down the data and store it on a MySQL database anyway, which basically means I had to code a script to automatically get the data from their web services and store it in our database every night. There is a huge amount of data in their systems and I only pull the data for four manufacturers.
So I built the scripts. I built the front end, even though certain things weren't documented, I managed to get them to finally send me some documentation on how to use their data effectively, and as of yesterday, the site is finally complete and finished.
Needless to say, it doesn't really look like it should have taken four or five weeks of work to produce the site, considering some other sites I've done, excluding the bespoke car price quoting system have taken two days at most. In fact, the actual site only took two days to build around the drupal module I made to interface with my back end data. As far as my boss is concerned, I've been sitting on my ass browsing HN and reddit for four weeks, and then took a week and a half to complete the site once I was bored with not working.
It has come to the point where whenever he starts his rant, I put my headphones on, or I shout at him, or I get very angry and walk off for a while. I really cannot work under these conditions for very much longer, and to be honest I am really ashamed with how I am handling his rants and outbursts.
I have built so many sites, systems for publishing stock to stock aggregates, systems for publishing stock to our own websites using different data we get back from the other dealerships, loyalty systems, niche seo websites, you name it, I've done it. I have made this man and this company a lot of money, and it's cost them a little over two years of my salary, which is paltry to say the least with how much I've worked.
They even outsourced me to develop a ticketing system for a [potentially recognisable car service] company, to the tune of more than half my annual salary for four weeks work.
It's an abusive environment, regardless of the fact that it's overlooked if I come in 20 minutes late in the morning, so long as I make it up at the end.
I'm in horrible position though, I really do like my boss in every other capacity other than a boss. His lack of understanding of what it is I do, and the fact that he can't seem to a) estimate how long a project will take me, and b) realise that not all programming work is instantly visible makes me think that I am going to quit in an epic fashion of diving off the balcony head first into an R8 Spyder and break my neck on the fucking windscreen.
Therein lies the problem. Where I live, there are no jobs. Some statistics recently suggested that a snowball had a better chance of opening an ice cream shop in the deepest layer of hell than I have of getting another job.
Although I have two years experience (more than double that if you count the freelancing I have done in a non salaried position) in PHP, MySQL, HTML, Javascript, LAMP Server administration, networking, etc, there aren't enough jobs to go around the unemployed in this city, let alone some guy with no degree.
Trying to discuss this with him really isn't working. Any advice?
TL;DR: Does anyone else browse the net in work when they are having a hard time focusing, but not blatantly trying to hide it and yet get rebuked even when the only other option is to sit, knuckles dragging drooling on the keyboard whilst staring at the screen because your brain no longer works?"},"title":{"matchLevel":"none","matchedWords":[],"value":"Ask HN: \"Is that work?\""}},"_tags":["story","author_throwaway_burn","story_1842719","ask_hn"],"author":"throwaway_burn","children":[1842764,1842792,1842797,1842800,1842816,1842840,1842862,1842865,1842872,1842888,1842889,1842904,1842909,1842923,1842924,1842943,1842966,1842971,1842988,1842992,1842997,1843005,1843032,1843100,1843115,1843116,1843147,1843254,1843266,1843275,1843301,1843355,1843439,1843508,1843535,1843724,1843773,1844811,1845025,1845238,1845807,1852862,1852937],"created_at":"2010-10-28T13:17:17Z","created_at_i":1288271837,"num_comments":105,"objectID":"1842719","points":83,"story_id":1842719,"story_text":"It's a commonly heard phrase in my office.
I'm the only developer where I work, based in a [car] dealership, which is one of the franchise locations of a car dealer group which owns [more than a few] (or so) other dealerships in quite a large area of the United Kingdom. For the record, I am not writing this on company time [half an hour lunch break, although he still does this even if I AM on my lunch break]
I build websites, web apps, microsites, soon to be Android Apps and iPhone apps. I am in charge of slicing and dicing photoshop templates which we get from our designer (Who sits behind me in the same office) into HTML templates to use with Drupal, CakePHP et al. I am in charge of inputting the copy for said microsites and websites from MS Word documents, this copy is written by the \"Marketing Manager\", who currently sits behind and to the right of me. To the right of where I sit, is my boss, the son of one of the Managing Directors of this company.
Do not get the tone of this post wrong, I love my job. I love the people, I love the fact that it's only twenty minutes down the road from my house, I love that parking is provided (although, not on site, about a quarter of a mile down the road). I even love my boss, he's a good, capable man and just wants to get things done and do what's best for this company.
However, as my work load has increased over the last two years, and things have gotten more difficult for me (harder problems, taking longer to solve and a lot more thinking) my brain has gotten tired. I've become burnt out maybe three times in the last two years, and have had to take a week off in some cases just to recover from the last project. (Considering we get 20 days holiday and zero sick pay, whenever I AM actually sick, I need to take holiday or I don't get paid for those days).
Sometimes, when I have really lost focus, or a problem has just circled in my head for too long to be solved without taking a break from thinking about it, I'll browse HN or reddit, not for very long, just for about 5 or 10 minutes until my head clears. More often than not, I'll be searching for the problem I have and reading articles about it.
My boss has the tremendous habit (when he is actually in the dealership), of coming into the office and looking straight at my screen, with the now dreaded words, \"Is that work?\", usually followed by \"Come on [throwaway_burn]! Focus! Get some fucking work done!\".
Now, my boss is a very visual person. If he can't instantly see results, he assumed I'm not doing any work. I've just spend four or five weeks creating a completely bespoke front end for building car quotes based on a completely experimental Web Service API built by a car valuation service. They usually make you host an MSSQL server, and they push or replicate their data to your server so you always have an up to date set of data to build cars with. My boss decided that this wasn't the way to go, because it's quite expensive so usually a single dealer group couldn't afford it, hence why you have aggregation systems, and people have it built into their customer management systems. [The web service company] probably lets us use this system for cheap as it is so undocumented when I first started using it, and it very slow, which means I have to pull down the data and store it on a MySQL database anyway, which basically means I had to code a script to automatically get the data from their web services and store it in our database every night. There is a huge amount of data in their systems and I only pull the data for four manufacturers.
So I built the scripts. I built the front end, even though certain things weren't documented, I managed to get them to finally send me some documentation on how to use their data effectively, and as of yesterday, the site is finally complete and finished.
Needless to say, it doesn't really look like it should have taken four or five weeks of work to produce the site, considering some other sites I've done, excluding the bespoke car price quoting system have taken two days at most. In fact, the actual site only took two days to build around the drupal module I made to interface with my back end data. As far as my boss is concerned, I've been sitting on my ass browsing HN and reddit for four weeks, and then took a week and a half to complete the site once I was bored with not working.
It has come to the point where whenever he starts his rant, I put my headphones on, or I shout at him, or I get very angry and walk off for a while. I really cannot work under these conditions for very much longer, and to be honest I am really ashamed with how I am handling his rants and outbursts.
I have built so many sites, systems for publishing stock to stock aggregates, systems for publishing stock to our own websites using different data we get back from the other dealerships, loyalty systems, niche seo websites, you name it, I've done it. I have made this man and this company a lot of money, and it's cost them a little over two years of my salary, which is paltry to say the least with how much I've worked.
They even outsourced me to develop a ticketing system for a [potentially recognisable car service] company, to the tune of more than half my annual salary for four weeks work.
It's an abusive environment, regardless of the fact that it's overlooked if I come in 20 minutes late in the morning, so long as I make it up at the end.
I'm in horrible position though, I really do like my boss in every other capacity other than a boss. His lack of understanding of what it is I do, and the fact that he can't seem to a) estimate how long a project will take me, and b) realise that not all programming work is instantly visible makes me think that I am going to quit in an epic fashion of diving off the balcony head first into an R8 Spyder and break my neck on the fucking windscreen.
Therein lies the problem. Where I live, there are no jobs. Some statistics recently suggested that a snowball had a better chance of opening an ice cream shop in the deepest layer of hell than I have of getting another job.
Although I have two years experience (more than double that if you count the freelancing I have done in a non salaried position) in PHP, MySQL, HTML, Javascript, LAMP Server administration, networking, etc, there aren't enough jobs to go around the unemployed in this city, let alone some guy with no degree.
Trying to discuss this with him really isn't working. Any advice?
TL;DR: Does anyone else browse the net in work when they are having a hard time focusing, but not blatantly trying to hide it and yet get rebuked even when the only other option is to sit, knuckles dragging drooling on the keyboard whilst staring at the screen because your brain no longer works?","title":"Ask HN: \"Is that work?\"","updated_at":"2026-03-25T04:40:22Z"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"sklaiber"},"story_text":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["seo","data","api","expensive"],"value":"I kept losing recipes. You know how it goes \u2014 you're scrolling TikTok at midnight, see an amazing pasta dish, save it, and never find it again. So I built TasteBuddy to fix that for myself. What I didn't expect: parsing recipes from the internet is a rabbit hole that goes deep.
The thing is, recipe content is scattered everywhere in completely different formats. A food blog might have nice JSON-LD markup. A TikTok? Just someone talking over a video. An Instagram reel? Recipe buried in the comments. Pinterest? Links to blogs that died three years ago.
So I ended up building specialized extractors for each platform.
*Websites* are the "easy" case. I look for JSON-LD with `@type: Recipe` first \u2014 most food blogs have it, thanks to SEO plugins. But the real world is messy. I've seen duration fields as `PT30M`, `30 minutes`, `0:30`, and my personal favorite, just `half an hour`. About 30% of recipe URLs have no structured data at all, so I fall back to Gemini to make sense of the raw HTML.
*TikTok* is where it gets fun. There's no recipe API. My pipeline resolves short URLs, then checks if the creator says something like "link in bio" (I detect this in five languages because German food TikTok is surprisingly massive). If I can find their website, great \u2014 I scrape the actual recipe from there. If not, I download the video via Apify and let Gemini analyze the frames. It works, but it's slow and expensive, so that's a Pro-only feature.
*Instagram and Facebook* \u2014 similar deal. oEmbed gets me the image, but the recipe is usually in the caption or comments. Same link-in-bio detection, same website resolution.
*Photos* are actually straightforward \u2014 screenshot of a recipe, photo of a cookbook page, whatever. Gemini's vision model handles those surprisingly well.
*One thing I'm proud of: the AI tiering.* Not every task needs a big model.
- Gemini Flash Lite handles 90% of the work \u2014 classifying content ("is this even a recipe?"), parsing ingredients, extracting recipe names from social media captions. Cheap, fast, good enough.\n- Gemini Flash kicks in when structured data fails \u2014 parsing messy HTML, analyzing video frames, processing social media posts.\n- Gemini Pro only for image generation (recipe share cards).\n- text-embedding-004 for semantic search across your recipe collection.
This keeps my costs sane as a solo dev. Using Flash for everything would've been 10x more expensive with barely better results for the simple tasks.
*Stuff I learned the hard way:*
- JSON-LD in the wild is chaos. The spec is fine, but WordPress plugins are creative.\n- "Link in bio" is how recipe distribution actually works on social media. Detecting that pattern is more valuable than trying to parse a video.\n- AI as fallback beats AI as default. Structured data first, AI when it fails = 95%+ success at a fraction of the cost.\n- Tier your models aggressively. Don't throw dollars at a problem that cents can solve.
*Stack:* Flutter (just me, indie dev), Supabase (Postgres + Deno Edge Functions), Gemini, Apify, PostHog.
Free with a Pro tier for video extraction and household sharing.
Happy to go deeper on any part of the extraction pipeline.
https://taste-buddy.app"},"title":{"matchLevel":"none","matchedWords":[],"value":"Show HN: I extract recipes from TikTok, Instagram, and the messy web"}},"_tags":["story","author_sklaiber","story_46974890","show_hn"],"author":"sklaiber","children":[46999993],"created_at":"2026-02-11T13:45:46Z","created_at_i":1770817546,"num_comments":0,"objectID":"46974890","points":3,"story_id":46974890,"story_text":"I kept losing recipes. You know how it goes \u2014 you're scrolling TikTok at midnight, see an amazing pasta dish, save it, and never find it again. So I built TasteBuddy to fix that for myself. What I didn't expect: parsing recipes from the internet is a rabbit hole that goes deep.
The thing is, recipe content is scattered everywhere in completely different formats. A food blog might have nice JSON-LD markup. A TikTok? Just someone talking over a video. An Instagram reel? Recipe buried in the comments. Pinterest? Links to blogs that died three years ago.
So I ended up building specialized extractors for each platform.
*Websites* are the "easy" case. I look for JSON-LD with `@type: Recipe` first \u2014 most food blogs have it, thanks to SEO plugins. But the real world is messy. I've seen duration fields as `PT30M`, `30 minutes`, `0:30`, and my personal favorite, just `half an hour`. About 30% of recipe URLs have no structured data at all, so I fall back to Gemini to make sense of the raw HTML.
*TikTok* is where it gets fun. There's no recipe API. My pipeline resolves short URLs, then checks if the creator says something like "link in bio" (I detect this in five languages because German food TikTok is surprisingly massive). If I can find their website, great \u2014 I scrape the actual recipe from there. If not, I download the video via Apify and let Gemini analyze the frames. It works, but it's slow and expensive, so that's a Pro-only feature.
*Instagram and Facebook* \u2014 similar deal. oEmbed gets me the image, but the recipe is usually in the caption or comments. Same link-in-bio detection, same website resolution.
*Photos* are actually straightforward \u2014 screenshot of a recipe, photo of a cookbook page, whatever. Gemini's vision model handles those surprisingly well.
*One thing I'm proud of: the AI tiering.* Not every task needs a big model.
- Gemini Flash Lite handles 90% of the work \u2014 classifying content ("is this even a recipe?"), parsing ingredients, extracting recipe names from social media captions. Cheap, fast, good enough.\n- Gemini Flash kicks in when structured data fails \u2014 parsing messy HTML, analyzing video frames, processing social media posts.\n- Gemini Pro only for image generation (recipe share cards).\n- text-embedding-004 for semantic search across your recipe collection.
This keeps my costs sane as a solo dev. Using Flash for everything would've been 10x more expensive with barely better results for the simple tasks.
*Stuff I learned the hard way:*
- JSON-LD in the wild is chaos. The spec is fine, but WordPress plugins are creative.\n- "Link in bio" is how recipe distribution actually works on social media. Detecting that pattern is more valuable than trying to parse a video.\n- AI as fallback beats AI as default. Structured data first, AI when it fails = 95%+ success at a fraction of the cost.\n- Tier your models aggressively. Don't throw dollars at a problem that cents can solve.
*Stack:* Flutter (just me, indie dev), Supabase (Postgres + Deno Edge Functions), Gemini, Apify, PostHog.
Free with a Pro tier for video extraction and household sharing.
Happy to go deeper on any part of the extraction pipeline.
https://taste-buddy.app","title":"Show HN: I extract recipes from TikTok, Instagram, and the messy web","updated_at":"2026-03-05T23:34:41Z"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"sklaiber"},"story_text":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["seo","data","api","expensive"],"value":"I kept losing recipes. You know how it goes \u2014 you're scrolling TikTok at midnight, see an amazing pasta dish, save it, and never find it again. So I built TasteBuddy to fix that for myself. What I didn't expect: parsing recipes from the internet is a rabbit hole that goes deep.
The thing is, recipe content is scattered everywhere in completely different formats. A food blog might have nice JSON-LD markup. A TikTok? Just someone talking over a video. An Instagram reel? Recipe buried in the comments. Pinterest? Links to blogs that died three years ago.
So I ended up building specialized extractors for each platform.
Websites are the "easy" case. I look for JSON-LD with @type: Recipe first \u2014 most food blogs have it, thanks to SEO plugins. But the real world is messy. I've seen duration fields as PT30M, 30 minutes, 0:30, and my personal favorite, just half an hour. About 30% of recipe URLs have no structured data at all, so I fall back to Gemini to make sense of the raw HTML.
TikTok is where it gets fun. There's no recipe API. My pipeline resolves short URLs, then checks if the creator says something like "link in bio" (I detect this in five languages because German food TikTok is surprisingly massive). If I can find their website, great \u2014 I scrape the actual recipe from there. If not, I download the video via Apify and let Gemini analyze the frames. It works, but it's slow and expensive, so that's a Pro-only feature.
Instagram and Facebook \u2014 similar deal. oEmbed gets me the image, but the recipe is usually in the caption or comments. Same link-in-bio detection, same website resolution.
Photos are actually straightforward \u2014 screenshot of a recipe, photo of a cookbook page, whatever. Gemini's vision model handles those surprisingly well.
One thing I'm proud of: the AI tiering. Not every task needs a big model.
- Gemini Flash Lite handles 90% of the work \u2014 classifying content, parsing ingredients, extracting recipe names. Cheap, fast, good enough.\n- Gemini Flash kicks in when structured data fails \u2014 parsing messy HTML, analyzing video frames.\n- Gemini Pro only for image generation (recipe share cards).\n- text-embedding-004 for semantic search across your recipe collection.
This keeps my costs sane as a solo dev.
Stuff I learned the hard way:
- JSON-LD in the wild is chaos. The spec is fine, but WordPress plugins are creative.\n- "Link in bio" is how recipe distribution actually works on social media.\n- AI as fallback beats AI as default. Structured data first, AI when it fails = 95%+ success at a fraction of the cost.\n- Tier your models aggressively. Don't throw dollars at a problem that cents can solve.
Stack: Flutter (just me, indie dev), Supabase (Postgres + Deno Edge Functions), Gemini, Apify, PostHog.
Free with a Pro tier for video extraction and household sharing. Happy to go deeper on any part of the extraction pipeline."},"title":{"matchLevel":"none","matchedWords":[],"value":"Show HN: I extract recipes from TikTok, Instagram, and the messy web"},"url":{"matchLevel":"none","matchedWords":[],"value":"https://taste-buddy.app"}},"_tags":["story","author_sklaiber","story_46974801","show_hn"],"author":"sklaiber","created_at":"2026-02-11T13:38:38Z","created_at_i":1770817118,"num_comments":0,"objectID":"46974801","points":1,"story_id":46974801,"story_text":"I kept losing recipes. You know how it goes \u2014 you're scrolling TikTok at midnight, see an amazing pasta dish, save it, and never find it again. So I built TasteBuddy to fix that for myself. What I didn't expect: parsing recipes from the internet is a rabbit hole that goes deep.
The thing is, recipe content is scattered everywhere in completely different formats. A food blog might have nice JSON-LD markup. A TikTok? Just someone talking over a video. An Instagram reel? Recipe buried in the comments. Pinterest? Links to blogs that died three years ago.
So I ended up building specialized extractors for each platform.
Websites are the "easy" case. I look for JSON-LD with @type: Recipe first \u2014 most food blogs have it, thanks to SEO plugins. But the real world is messy. I've seen duration fields as PT30M, 30 minutes, 0:30, and my personal favorite, just half an hour. About 30% of recipe URLs have no structured data at all, so I fall back to Gemini to make sense of the raw HTML.
TikTok is where it gets fun. There's no recipe API. My pipeline resolves short URLs, then checks if the creator says something like "link in bio" (I detect this in five languages because German food TikTok is surprisingly massive). If I can find their website, great \u2014 I scrape the actual recipe from there. If not, I download the video via Apify and let Gemini analyze the frames. It works, but it's slow and expensive, so that's a Pro-only feature.
Instagram and Facebook \u2014 similar deal. oEmbed gets me the image, but the recipe is usually in the caption or comments. Same link-in-bio detection, same website resolution.
Photos are actually straightforward \u2014 screenshot of a recipe, photo of a cookbook page, whatever. Gemini's vision model handles those surprisingly well.
One thing I'm proud of: the AI tiering. Not every task needs a big model.
- Gemini Flash Lite handles 90% of the work \u2014 classifying content, parsing ingredients, extracting recipe names. Cheap, fast, good enough.\n- Gemini Flash kicks in when structured data fails \u2014 parsing messy HTML, analyzing video frames.\n- Gemini Pro only for image generation (recipe share cards).\n- text-embedding-004 for semantic search across your recipe collection.
This keeps my costs sane as a solo dev.
Stuff I learned the hard way:
- JSON-LD in the wild is chaos. The spec is fine, but WordPress plugins are creative.\n- "Link in bio" is how recipe distribution actually works on social media.\n- AI as fallback beats AI as default. Structured data first, AI when it fails = 95%+ success at a fraction of the cost.\n- Tier your models aggressively. Don't throw dollars at a problem that cents can solve.
Stack: Flutter (just me, indie dev), Supabase (Postgres + Deno Edge Functions), Gemini, Apify, PostHog.
Free with a Pro tier for video extraction and household sharing. Happy to go deeper on any part of the extraction pipeline.","title":"Show HN: I extract recipes from TikTok, Instagram, and the messy web","updated_at":"2026-03-05T23:34:41Z","url":"https://taste-buddy.app"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"hawky89"},"comment_text":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["seo","data","api","expensive"],"value":"Hey HN,
For my WordPress blog, I was constantly frustrated by how expensive good backlink prospecting tools are. So, I decided to build my own.
It's a simple tool that helps identify potential English backlink opportunities for WordPress blogs based on keywords. The interesting part is it runs entirely on the free tiers of the Brave Search API and Google Gemini API, meaning no hefty monthly subscriptions for the data itself.
It's available as a small Gumroad product. I've focused on keeping it lean and effective for anyone looking to boost their SEO without breaking the bank.
Happy to answer any questions about the technical approach or how it works!"},"story_title":{"matchLevel":"none","matchedWords":[],"value":"Built a small tool to find English backlink opportunities for WordPress blogs"},"story_url":{"matchLevel":"none","matchedWords":[],"value":"https://einkommensquellen.gumroad.com/l/Search_Backlinks"}},"_tags":["comment","author_hawky89","story_44051511"],"author":"hawky89","comment_text":"Hey HN,
For my WordPress blog, I was constantly frustrated by how expensive good backlink prospecting tools are. So, I decided to build my own.
It's a simple tool that helps identify potential English backlink opportunities for WordPress blogs based on keywords. The interesting part is it runs entirely on the free tiers of the Brave Search API and Google Gemini API, meaning no hefty monthly subscriptions for the data itself.
It's available as a small Gumroad product. I've focused on keeping it lean and effective for anyone looking to boost their SEO without breaking the bank.
Happy to answer any questions about the technical approach or how it works!","created_at":"2025-05-21T13:56:47Z","created_at_i":1747835807,"objectID":"44051512","parent_id":44051511,"story_id":44051511,"story_title":"Built a small tool to find English backlink opportunities for WordPress blogs","story_url":"https://einkommensquellen.gumroad.com/l/Search_Backlinks","updated_at":"2025-05-21T13:57:55Z"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"johnerik"},"comment_text":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["seo","data","api","expensive"],"value":"Hi HN! We built a MacBook Upgrade Program inspired by Apple\u2019s iPhone Upgrade Program, so you can finance a new MacBook (with AppleCare+) over 36 months\u2014and upgrade after 24 months if you want the latest model. It\u2019s live at getupgraded.com (US only for now).
I\u2019m @johnerik the cofounder/ceo and I\u2019ve been running upgrade programs for other retailers in Scandinavia for the last 8 years. Now we\u2019re building our own! I just moved (back) to Austin.
## Why?
- Because we\u2019ve wanted a MacBook Upgrade Program since Apple launched the iPhone Upgrade Program, in 2015! (I personally have been on the iPhone program since it launched. I\u2019ll admit that I have skipped some years and not upgraded, but have been loyal to the program.)
- Because the whole process of buying a MacBook in cash and selling on Facebook Marketplace when you\u2019re ready to upgrade is a hassle. (I would end up with devices laying around because I disliked the selling part so much.)
- Because if we can get this flywheel started \u2013 where we get devices back frequently, repair them when needed and then quickly find them a new home \u2013 I think we can increase the useful life of a given device, and therefore reduce the total number of devices that need to be produced overall.
- Because as a team we have a ton of experience doing this. Since 2016 we\u2019ve been running upgrade programs for retailers and brands (including Apple and Microsoft) in Scandinavia.
- Because we think there\u2019s an opportunity to build the next great American electronics retailer. New MacBooks are just the start. Used and refurbished devices are the future.
## How does this work?
- The program divides the total cost of a new MacBook and AppleCare+ into 36 monthly payments.
- After making 24 monthly payments (two-thirds of the total), you're eligible to upgrade your MacBook.
- To upgrade, you\u2019d just choose your new MacBook, and we'll ship it to you right away.
- We'll temporarily pause all payments for 30 days while you have both the old and new MacBook.
- You'll transfer your data from the old MacBook to the new one, then send the old one back to us using the provided prepaid return box. (We\u2019re considering including a Thunderbolt cable to make migration super fast \u2013 love to hear if you think this would be useful.)
- After we verify your old MacBook is in good working condition (normal wear and tear is acceptable and expected), we'll pay off the remaining 12 payments (the final one-third) for you.
- You'll then start fresh with a new set of 36 monthly payments on your upgraded MacBook.
If you don\u2019t want to upgrade, no prob. Just make all 36 payments (pay off all the principal) and you\u2019re done.
## How do the economics for this come together?
Our main partners are
1. Gatortec, an Apple Authorized Reseller, \n2. Citizen Bank, also the lender for Apple\u2019s iUp Program, and \n3. A bulk buyer for the used devices.
- Gatortec has a direct line to purchase Apple devices/services below MSRP and they share the majority of the margin with us.
- Citizens gives our 740+ credit customers 0% interest and 640-739 customers a variable rate between 29.99 and 12.99%. In exchange, we give them half our margin. And keep the rest as our upfront take rate.
- Our bulk buyer has guaranteed to buy the used MacBooks for 37% of MSRP. We are buying them from customers for ~33% of MSRP (it\u2019s actually slightly higher on loans with interest because of how amortization works, but I\u2019ll skip that here). So we theoretically will make that spread. Though our actual plan is to refurb these devices ourselves, resell them again on our marketplace, and give as much cash back to customers as possible \u2013 keeping just enough to cover our expenses.
In summary: We make a little bit from the purchase itself, and a little bit more from the upgrade. The goal is loyal customers, and therefore a higher lifetime value.
## What\u2019s our tech stack?
- Ruby on Rails backend, because it\u2019s what we knew at founding and we wanted to move fast.
- No fancy JS framework or SPA. Just HTML, CSS, and JavaScript (leveraging Rails tooling).
- Heroku hosting, because it Just Works.
- Some AWS (Lambda, API Gateway, DNS, CDN, S3, ...) because Heroku doesn't always Just Work (especially with things that need to scale).
- SaaS tools like Sentry, Algolia, Posthog, Mailgun, \u2026
- Tons of great open source libraries (and we contribute where we can, especially with bugs or features). Again, try not to build stuff.
- We've settled on Windsurf as our AI-driven IDE, but Cursor is also great. The space is moving so fast so we might be using something else next month (but we try not to switch too often).
- We have a large, multi-regional, internationalized codebase with 10-years of history, so we're constantly refactoring, cleaning, pruning, updating (linters, tests... this part is still not easy to hand off to AI, but maybe soon!).
- A Wordpress marketing site (:vomit:... not fun to work with or maintain) because someone said it's the best for SEO, but we\u2019ll likely rebuild and host in Framer soon.
- We maintain our retailers' Shopify store, and we can yarn dev the store, but it's clunky to have so many moving parts that don't run on our Macs (airplane mode FTW).
## Where we are going
- We want to create a new kind of online electronics retailer. One that has A+ customer support, sells both new and used devices, serves both consumers and businesses, and has various ways to buy and sell.
- Long term, we want to be a one stop shop for managing all your devices. Families are like small businesses now, both have 20 devices! We\u2019d want to build an app where you can track all your devices, see when certain ones are due for an upgrade, compare prices, and more. Love any feedback here.
- I\u2019ll also just say that for me personally, I know we are selling brand new MacBooks right now and that\u2019s super high-end, but for now we\u2019re just starting with the most expensive device and we will work our way down. This whole idea of taking care of devices, fixing them, and using them again is very much the vibe I grew up with. My dad is a mechanic so we were way more likely to repair something than we were to buy something new.
## Finally, I\u2019d love your feedback/questions on anything above, or here are specific questions:
- User experience/design: How clear is our flow? Is the site layout friendly and accessible? What questions are not answered? We know it\u2019s super Apple-like.
- Features: Any must-have features you\u2019d expect from a retailer/upgrade program?
- Financing vs leasing vs subscriptions: The underlying financial instrument today is a term loan, but we kind of squeezing something that\u2019s similar to a lease or a subscription into a loan.
- B2B: We\u2019ve tested leasing to a few friends' companies? Would you want this for your company?
- Leanings from other countries: We\u2019re currently US-only\u2014is there anything cool in your country/locale that we could learn from?
- Feel free to test our customer support chat.
- Future product categories: Which devices or brands should we consider adding next?
- Circular Economy: Would you want to know your COe savings are? Or is that just greenwashing at this point.
THANK YOU!"},"story_title":{"matchLevel":"none","matchedWords":[],"value":"Show HN: Upgraded \u2013 A MacBook Upgrade Program"},"story_url":{"matchLevel":"none","matchedWords":[],"value":"https://getupgraded.com/"}},"_tags":["comment","author_johnerik","story_43561492"],"author":"johnerik","children":[43561573],"comment_text":"Hi HN! We built a MacBook Upgrade Program inspired by Apple\u2019s iPhone Upgrade Program, so you can finance a new MacBook (with AppleCare+) over 36 months\u2014and upgrade after 24 months if you want the latest model. It\u2019s live at getupgraded.com (US only for now).
I\u2019m @johnerik the cofounder/ceo and I\u2019ve been running upgrade programs for other retailers in Scandinavia for the last 8 years. Now we\u2019re building our own! I just moved (back) to Austin.
## Why?
- Because we\u2019ve wanted a MacBook Upgrade Program since Apple launched the iPhone Upgrade Program, in 2015! (I personally have been on the iPhone program since it launched. I\u2019ll admit that I have skipped some years and not upgraded, but have been loyal to the program.)
- Because the whole process of buying a MacBook in cash and selling on Facebook Marketplace when you\u2019re ready to upgrade is a hassle. (I would end up with devices laying around because I disliked the selling part so much.)
- Because if we can get this flywheel started \u2013 where we get devices back frequently, repair them when needed and then quickly find them a new home \u2013 I think we can increase the useful life of a given device, and therefore reduce the total number of devices that need to be produced overall.
- Because as a team we have a ton of experience doing this. Since 2016 we\u2019ve been running upgrade programs for retailers and brands (including Apple and Microsoft) in Scandinavia.
- Because we think there\u2019s an opportunity to build the next great American electronics retailer. New MacBooks are just the start. Used and refurbished devices are the future.
## How does this work?
- The program divides the total cost of a new MacBook and AppleCare+ into 36 monthly payments.
- After making 24 monthly payments (two-thirds of the total), you're eligible to upgrade your MacBook.
- To upgrade, you\u2019d just choose your new MacBook, and we'll ship it to you right away.
- We'll temporarily pause all payments for 30 days while you have both the old and new MacBook.
- You'll transfer your data from the old MacBook to the new one, then send the old one back to us using the provided prepaid return box. (We\u2019re considering including a Thunderbolt cable to make migration super fast \u2013 love to hear if you think this would be useful.)
- After we verify your old MacBook is in good working condition (normal wear and tear is acceptable and expected), we'll pay off the remaining 12 payments (the final one-third) for you.
- You'll then start fresh with a new set of 36 monthly payments on your upgraded MacBook.
If you don\u2019t want to upgrade, no prob. Just make all 36 payments (pay off all the principal) and you\u2019re done.
## How do the economics for this come together?
Our main partners are
1. Gatortec, an Apple Authorized Reseller, \n2. Citizen Bank, also the lender for Apple\u2019s iUp Program, and \n3. A bulk buyer for the used devices.
- Gatortec has a direct line to purchase Apple devices/services below MSRP and they share the majority of the margin with us.
- Citizens gives our 740+ credit customers 0% interest and 640-739 customers a variable rate between 29.99 and 12.99%. In exchange, we give them half our margin. And keep the rest as our upfront take rate.
- Our bulk buyer has guaranteed to buy the used MacBooks for 37% of MSRP. We are buying them from customers for ~33% of MSRP (it\u2019s actually slightly higher on loans with interest because of how amortization works, but I\u2019ll skip that here). So we theoretically will make that spread. Though our actual plan is to refurb these devices ourselves, resell them again on our marketplace, and give as much cash back to customers as possible \u2013 keeping just enough to cover our expenses.
In summary: We make a little bit from the purchase itself, and a little bit more from the upgrade. The goal is loyal customers, and therefore a higher lifetime value.
## What\u2019s our tech stack?
- Ruby on Rails backend, because it\u2019s what we knew at founding and we wanted to move fast.
- No fancy JS framework or SPA. Just HTML, CSS, and JavaScript (leveraging Rails tooling).
- Heroku hosting, because it Just Works.
- Some AWS (Lambda, API Gateway, DNS, CDN, S3, ...) because Heroku doesn't always Just Work (especially with things that need to scale).
- SaaS tools like Sentry, Algolia, Posthog, Mailgun, \u2026
- Tons of great open source libraries (and we contribute where we can, especially with bugs or features). Again, try not to build stuff.
- We've settled on Windsurf as our AI-driven IDE, but Cursor is also great. The space is moving so fast so we might be using something else next month (but we try not to switch too often).
- We have a large, multi-regional, internationalized codebase with 10-years of history, so we're constantly refactoring, cleaning, pruning, updating (linters, tests... this part is still not easy to hand off to AI, but maybe soon!).
- A Wordpress marketing site (:vomit:... not fun to work with or maintain) because someone said it's the best for SEO, but we\u2019ll likely rebuild and host in Framer soon.
- We maintain our retailers' Shopify store, and we can yarn dev the store, but it's clunky to have so many moving parts that don't run on our Macs (airplane mode FTW).
## Where we are going
- We want to create a new kind of online electronics retailer. One that has A+ customer support, sells both new and used devices, serves both consumers and businesses, and has various ways to buy and sell.
- Long term, we want to be a one stop shop for managing all your devices. Families are like small businesses now, both have 20 devices! We\u2019d want to build an app where you can track all your devices, see when certain ones are due for an upgrade, compare prices, and more. Love any feedback here.
- I\u2019ll also just say that for me personally, I know we are selling brand new MacBooks right now and that\u2019s super high-end, but for now we\u2019re just starting with the most expensive device and we will work our way down. This whole idea of taking care of devices, fixing them, and using them again is very much the vibe I grew up with. My dad is a mechanic so we were way more likely to repair something than we were to buy something new.
## Finally, I\u2019d love your feedback/questions on anything above, or here are specific questions:
- User experience/design: How clear is our flow? Is the site layout friendly and accessible? What questions are not answered? We know it\u2019s super Apple-like.
- Features: Any must-have features you\u2019d expect from a retailer/upgrade program?
- Financing vs leasing vs subscriptions: The underlying financial instrument today is a term loan, but we kind of squeezing something that\u2019s similar to a lease or a subscription into a loan.
- B2B: We\u2019ve tested leasing to a few friends' companies? Would you want this for your company?
- Leanings from other countries: We\u2019re currently US-only\u2014is there anything cool in your country/locale that we could learn from?
- Feel free to test our customer support chat.
- Future product categories: Which devices or brands should we consider adding next?
- Circular Economy: Would you want to know your COe savings are? Or is that just greenwashing at this point.
THANK YOU!","created_at":"2025-04-02T20:59:13Z","created_at_i":1743627553,"objectID":"43561498","parent_id":43561492,"story_id":43561492,"story_title":"Show HN: Upgraded \u2013 A MacBook Upgrade Program","story_url":"https://getupgraded.com/","updated_at":"2025-04-02T21:26:41Z"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"ejboustany"},"comment_text":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["seo","data","api","expensive"],"value":"When I launched on February 17, 2024, the market was flooded with AI-powered website builders. But here\u2019s the thing \u2014 most of these tools weren\u2019t offering anything you couldn\u2019t already do with WordPress, WIX, or Squarespace and I wouldn\u2019t use them even if they were free.
As a Software Engineer that is passionate and hungry about turning ideas into reality, I wanted to create something different. Not just another website builder, but a platform that truly helps founders (non-tech founders) to build their SaaS dreams.
At its core, PagePalooza gives founders complete control over their website\u2019s front end \u2014 you can add pages, sections, change colors, text, and images. But that\u2019s just the foundation and nothing special here. The real magic happens with our \u201cCustom Task Manager.\u201d
Think of it as your personal engineering team on demand. Need a custom sign-up system that reads user data from Excel? Want to integrate a unique feature? Instead of paying monthly subscriptions or hiring expensive developers, you can order these features as one-time tasks, built directly into your site by Palooza engineers.
The best part? You own all your custom APIs and database. No vendor lock-in, no limitations. Own the code and scale one feature at a time.
I knew that it is not easy to trust an engineer or a platform with building your app, I wanted to let founders be able to test the waters before diving deep. That\u2019s why we\u2019ve made it easy to start small. Many founders begin with simple custom features like that Excel-integrated signup form and a small dashboard which costs you a very small fraction and use it as an MVP.
Since launch, we\u2019ve welcomed over 3,200 visitors (according to Google Analytics) without spending a dollar on paid advertising. Our first paid client came in mid-June 2024, and that month peaked with about 300 visitors in a single week. All this growth came through organic SEO, consistent blogging, and daily social media content across YouTube, Instagram, and TikTok.
The platform is built with .Net Core and Angular.
I\u2019m grateful to have worked with amazing founders in building their web applications, and I\u2019ll be sharing their stories after the holidays (hopefully sending some traffic their way too!).
To all the builders out there: keep learning, keep adjusting, and stay consistent. Here\u2019s to a better year ahead!"},"story_title":{"matchLevel":"none","matchedWords":[],"value":"[dead]"}},"_tags":["comment","author_ejboustany","story_42660326"],"author":"ejboustany","comment_text":"When I launched on February 17, 2024, the market was flooded with AI-powered website builders. But here\u2019s the thing \u2014 most of these tools weren\u2019t offering anything you couldn\u2019t already do with WordPress, WIX, or Squarespace and I wouldn\u2019t use them even if they were free.
As a Software Engineer that is passionate and hungry about turning ideas into reality, I wanted to create something different. Not just another website builder, but a platform that truly helps founders (non-tech founders) to build their SaaS dreams.
At its core, PagePalooza gives founders complete control over their website\u2019s front end \u2014 you can add pages, sections, change colors, text, and images. But that\u2019s just the foundation and nothing special here. The real magic happens with our \u201cCustom Task Manager.\u201d
Think of it as your personal engineering team on demand. Need a custom sign-up system that reads user data from Excel? Want to integrate a unique feature? Instead of paying monthly subscriptions or hiring expensive developers, you can order these features as one-time tasks, built directly into your site by Palooza engineers.
The best part? You own all your custom APIs and database. No vendor lock-in, no limitations. Own the code and scale one feature at a time.
I knew that it is not easy to trust an engineer or a platform with building your app, I wanted to let founders be able to test the waters before diving deep. That\u2019s why we\u2019ve made it easy to start small. Many founders begin with simple custom features like that Excel-integrated signup form and a small dashboard which costs you a very small fraction and use it as an MVP.
Since launch, we\u2019ve welcomed over 3,200 visitors (according to Google Analytics) without spending a dollar on paid advertising. Our first paid client came in mid-June 2024, and that month peaked with about 300 visitors in a single week. All this growth came through organic SEO, consistent blogging, and daily social media content across YouTube, Instagram, and TikTok.
The platform is built with .Net Core and Angular.
I\u2019m grateful to have worked with amazing founders in building their web applications, and I\u2019ll be sharing their stories after the holidays (hopefully sending some traffic their way too!).
To all the builders out there: keep learning, keep adjusting, and stay consistent. Here\u2019s to a better year ahead!","created_at":"2025-01-10T21:15:58Z","created_at_i":1736543758,"objectID":"42660327","parent_id":42660326,"story_id":42660326,"story_title":"[dead]","updated_at":"2025-01-10T21:19:40Z"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"derefr"},"comment_text":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["seo","data","api","expensive"],"value":"1. People don't generally use the (big, whole-web-corpus-trained) general-purpose LLM base-models to generate bot slop for the web. Paying per API call to generate that kind of stuff would be far too expensive; it'd be like paying for eStamps to send spam email. Spambot developers use smaller open-source models, trained on much smaller corpuses, sized and quantized to generate text that's "just good enough" to pass muster. This creates a sampling bias in the word-associational "knowledge" the model is working from when generating.
2. Given how LLMs work, a prompt is a bias \u2014 they're one-and-the-same. You can't ask an LLM to write you a mystery novel without it somewhat adopting the writing quirks common to the particular mystery novels it has "read." Even the writing style you use in your prompt influences this bias. (It's common advice among "AI character" chatbot authors, to write the "character card" describing a character, in the style that you want the character speaking in, for exactly this reason.) Whatever prompt the developer uses, is going to bias the bot away from the statistical norm, toward the writing-style elements that exist within whatever hypersphere of association-space contains plausible completions of the prompt.
3. Bot authors do SEO too! They take the tf-idf metrics and keyword stuffing, and turn it into training data to fine-tune models, in effect creating "automated SEO experts" that write in the SEO-compatible style by default. (And in so doing, they introduce unintentional further bias, given that the SEO-optimized training dataset likely is not an otherwise-perfect representative sampling of writing style for the target language.)"},"story_title":{"matchLevel":"none","matchedWords":[],"value":"Why wordfreq will not be updated"},"story_url":{"matchLevel":"none","matchedWords":[],"value":"https://github.com/rspeer/wordfreq/blob/master/SUNSET.md"}},"_tags":["comment","author_derefr","story_41578483"],"author":"derefr","children":[41588658],"comment_text":"1. People don't generally use the (big, whole-web-corpus-trained) general-purpose LLM base-models to generate bot slop for the web. Paying per API call to generate that kind of stuff would be far too expensive; it'd be like paying for eStamps to send spam email. Spambot developers use smaller open-source models, trained on much smaller corpuses, sized and quantized to generate text that's "just good enough" to pass muster. This creates a sampling bias in the word-associational "knowledge" the model is working from when generating.
2. Given how LLMs work, a prompt is a bias \u2014 they're one-and-the-same. You can't ask an LLM to write you a mystery novel without it somewhat adopting the writing quirks common to the particular mystery novels it has "read." Even the writing style you use in your prompt influences this bias. (It's common advice among "AI character" chatbot authors, to write the "character card" describing a character, in the style that you want the character speaking in, for exactly this reason.) Whatever prompt the developer uses, is going to bias the bot away from the statistical norm, toward the writing-style elements that exist within whatever hypersphere of association-space contains plausible completions of the prompt.
3. Bot authors do SEO too! They take the tf-idf metrics and keyword stuffing, and turn it into training data to fine-tune models, in effect creating "automated SEO experts" that write in the SEO-compatible style by default. (And in so doing, they introduce unintentional further bias, given that the SEO-optimized training dataset likely is not an otherwise-perfect representative sampling of writing style for the target language.)","created_at":"2024-09-18T18:50:13Z","created_at_i":1726685413,"objectID":"41583935","parent_id":41583598,"story_id":41578483,"story_title":"Why wordfreq will not be updated","story_url":"https://github.com/rspeer/wordfreq/blob/master/SUNSET.md","updated_at":"2024-10-07T03:12:31Z"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"lifeisstillgood"},"comment_text":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["seo","data","api","expensive"],"value":"The airlines have to try. Flight safety is expensive.
I mean you can conceive of solutions:
* Why does the luggage not get treated like container transport. Load up cages next to check in desks for the same destination.
* Take back control over your booking. There is absolutely no need for any airline to not take bookings direct. Want to sell through your partners - fine but they come to your API/site to actually book it. Then everyone has real time info. Think that will drive your prices down? No, don't sell at loss making prices.
I could go on. I am having a hell just trying to find somewhere nice and sunny to take 2 children. There is just so many options and so little information to trust.
But this problems exists for every service - hair cutting, selling houses and airlines. One and only one URI is really selling an actual seat or an actual house or an actual chair with a hair dresser.
So one and only one URI should publish what is available.
And then aggregation and search is simple(ish)
It seems to me that the next evolution of the search engine is to tell the people selling stuff that they don't need to do SEO, or marketing. They just need to use a standard micro data / format for whatever they are selling.
A long time ago I worked for an ISP that gave 1MB of personal disk space free to everyone. We bumped to to 10 then 100 when we realised no one actually used it.
This failure to put our own stuff online is the cause of huge problems - real estate agents, facebook, AOL, every walled garden, every agent, all stem from people and businesses not publishing their own needs.
Must stop ranting now"},"story_title":{"matchLevel":"none","matchedWords":[],"value":"Somebody please, for the love of god, fix shipping/couriers"},"story_url":{"matchLevel":"none","matchedWords":[],"value":""}},"_tags":["comment","author_lifeisstillgood","story_3971212"],"author":"lifeisstillgood","children":[3971923],"comment_text":"The airlines have to try. Flight safety is expensive.
I mean you can conceive of solutions:
* Why does the luggage not get treated like container transport. Load up cages next to check in desks for the same destination.
* Take back control over your booking. There is absolutely no need for any airline to not take bookings direct. Want to sell through your partners - fine but they come to your API/site to actually book it. Then everyone has real time info. Think that will drive your prices down? No, don't sell at loss making prices.
I could go on. I am having a hell just trying to find somewhere nice and sunny to take 2 children. There is just so many options and so little information to trust.
But this problems exists for every service - hair cutting, selling houses and airlines. One and only one URI is really selling an actual seat or an actual house or an actual chair with a hair dresser.
So one and only one URI should publish what is available.
And then aggregation and search is simple(ish)
It seems to me that the next evolution of the search engine is to tell the people selling stuff that they don't need to do SEO, or marketing. They just need to use a standard micro data / format for whatever they are selling.
A long time ago I worked for an ISP that gave 1MB of personal disk space free to everyone. We bumped to to 10 then 100 when we realised no one actually used it.
This failure to put our own stuff online is the cause of huge problems - real estate agents, facebook, AOL, every walled garden, every agent, all stem from people and businesses not publishing their own needs.
Must stop ranting now","created_at":"2012-05-14T16:44:16Z","created_at_i":1337013856,"objectID":"3971832","parent_id":3971767,"story_id":3971212,"story_title":"Somebody please, for the love of god, fix shipping/couriers","story_url":"","updated_at":"2024-09-19T18:37:42Z"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"tomkin"},"comment_text":{"fullyHighlighted":false,"matchLevel":"partial","matchedWords":["api"],"value":"I think this is going to be the Photoshop where people seriously start considering a web-based alternative. Not because CS6 will be poor software, but because I think the division among users who use Photoshop for photo editing/professional photography/restoration/stationary/brochure design and those who use it for designing web sites and applications is getting greater. The latter needing less of the options available in Photoshop today, let alone CS6. I say this with over 10 years of experience as a designer/developer \u2013 Photoshop needs a Basecamp simplicity-inspired derivative. I think as we move to more institutional, common & predictable UI (if we do), we'll find that we're doing very simple tasks and that those tasks can be taken from Photoshop and automated for the purpose of web/app design and execution. I know there are a few alternatives to Photoshop already, but of all the ones I've seen (yes, even Pixelmator) none of them really go for just the web industry. I think the market for such a tool (hopefully web-based) would be pretty great at this point \u2013 and given the audience \u2013 using state-of-the-art browser API wouldn't be much of an issue.
Long ago, Adobe had this product called \"Photoshop Server\" or something similar. I believe it was based off of Macromedia's Generator (which generated server-side Flash content). The concept was neat - place variables within a PSD, upload it and allow the server to render out based on what you wanted to show. The problem with this, of course, was that it was proprietary, heavy, expensive and overall pretty crappy. It would be really interesting to see someone create a web-based web design/development application that interacted with some OSS library that most hosting companies would allow (like gd, ning, etc), read .json and interpret data within the image.
Probably a bad idea, but how big would a .json-based image be compared to the average jpg? Pretty big, I'll bet. But I wonder with gzip compression how manageable could it be? Would be really cool to define information (text inside of the image, accessibility, SEO capabilities) within a .json-based image. No license required and it would be readable by any human or software.
{\n \"dimensions\": { \"width\": 3, \"height\": 3 },\n \"data\": [\n [\"rgba(255,255,255,1)\",\"rgba(255,255,255,1)\",\"rgba(255,255,255,1)\"],\n [\"rgba(255,255,255,1)\",\"rgba(255,255,255,1)\",\"rgba(255,255,255,1)\"],\n [\"rgba(255,255,255,1)\",\"rgba(255,255,255,1)\",\"rgba(255,255,255,1)\"]\n ]\n }\nLong ago, Adobe had this product called \"Photoshop Server\" or something similar. I believe it was based off of Macromedia's Generator (which generated server-side Flash content). The concept was neat - place variables within a PSD, upload it and allow the server to render out based on what you wanted to show. The problem with this, of course, was that it was proprietary, heavy, expensive and overall pretty crappy. It would be really interesting to see someone create a web-based web design/development application that interacted with some OSS library that most hosting companies would allow (like gd, ning, etc), read .json and interpret data within the image.
Probably a bad idea, but how big would a .json-based image be compared to the average jpg? Pretty big, I'll bet. But I wonder with gzip compression how manageable could it be? Would be really cool to define information (text inside of the image, accessibility, SEO capabilities) within a .json-based image. No license required and it would be readable by any human or software.
{\n \"dimensions\": { \"width\": 3, \"height\": 3 },\n \"data\": [\n [\"rgba(255,255,255,1)\",\"rgba(255,255,255,1)\",\"rgba(255,255,255,1)\"],\n [\"rgba(255,255,255,1)\",\"rgba(255,255,255,1)\",\"rgba(255,255,255,1)\"],\n [\"rgba(255,255,255,1)\",\"rgba(255,255,255,1)\",\"rgba(255,255,255,1)\"]\n ]\n }\nLots of really good stuff going on here, a few bits of feedback, ideas, etc, and also responding to some of the comments in other threads.
* The quick explanation of how this works is pretty strong, but I think the differentiator/value over other services maybe isn't called out as well. To me, this seems pretty clearly ideal for smaller teams all in on JAMstack/serverless with a lean stack that don't/won't have an easy place to create an automated process for synchronization, and are more likely to be using a planetscale/supabase/neon where this model is more attractive. I would suggest adding some of that info, not only to help target customers better recognize the value prop, but also help discovery via SEO.
* While the landing page gives a good overview, as a dev, before I sign up, I want more details of how it works, more complete examples, etc. For SaaS apps targeting the general market, after landing/product pages,the most visited pages are generally use cases, success stories, industry specific info, etc. But from my experience and talking with many others, for dev audience, docs are the next thing people visit (if they aren't ready to sign up). You don't need every doc at once, but things like a quick start guide, concept overview, feature overview, etc can all be high value docs that you generally need anyways for your first customers
* With docs, you can better address and explain the security side of things. I would do less on the landing page in regards to security and push that to more in depth detail in the docs. Talk about best practices of creating a read-only user. Create guides for most popular db vendors, etc
* As has already been mentioned, you will get concerns about any access to the db and "why not an API". I think these commenters are right that this will be a deal breaker for many companies, but I don't agree that you want an API. Not only because I think that reduces some of the value prop but also because then you would need to store customer data. I obviously don't know what you are storing today, but I would think that this model could have a big advantage if you didn't need to store any PII data in cc.dev. The complexity of dealing with compliance and regulatory requirements is no small part of why something like MailChimp is expensive is that burden. I don't know if this is something you currently consider a value prop (or if you have engineered to support this)... But I certainly would :)
* To address the reality of private dbs/giving access to a db, I would look at potentially implementing an agent. This agent would then run the queries (still provided by the user) and just-in-time deliver the results when needed. Getting the model of this right that works across different companies view of security will mean this is probably best to do with a handful of potential customers.
* As far as deployment and monetization model... I would only open source it if you are giving up on commercialization or if you can open source a subset that can be useful enough to build a community. Once again, returning to JAMstack, maybe solve in open source (but integrate) some problem unique to those teams. As far as a paid self-hosting, given that this seems like a one man show, I would resist it. Trying to do SaaS and support of self-hosting is rough, even for well funded VC backed teams..
Quite a bit here, but if you want follows up on anything, my email is in profile or will reply to comments.
Congrats again and hope it goes somewhere :)"},"story_title":{"matchLevel":"none","matchedWords":[],"value":"Show HN: I made a MailChimp alternative that connects to your database"},"story_url":{"matchLevel":"none","matchedWords":[],"value":"https://cc.dev"}},"_tags":["comment","author_addisonj","story_36836596"],"author":"addisonj","children":[36838182],"comment_text":"Congrats on the launch!
Lots of really good stuff going on here, a few bits of feedback, ideas, etc, and also responding to some of the comments in other threads.
* The quick explanation of how this works is pretty strong, but I think the differentiator/value over other services maybe isn't called out as well. To me, this seems pretty clearly ideal for smaller teams all in on JAMstack/serverless with a lean stack that don't/won't have an easy place to create an automated process for synchronization, and are more likely to be using a planetscale/supabase/neon where this model is more attractive. I would suggest adding some of that info, not only to help target customers better recognize the value prop, but also help discovery via SEO.
* While the landing page gives a good overview, as a dev, before I sign up, I want more details of how it works, more complete examples, etc. For SaaS apps targeting the general market, after landing/product pages,the most visited pages are generally use cases, success stories, industry specific info, etc. But from my experience and talking with many others, for dev audience, docs are the next thing people visit (if they aren't ready to sign up). You don't need every doc at once, but things like a quick start guide, concept overview, feature overview, etc can all be high value docs that you generally need anyways for your first customers
* With docs, you can better address and explain the security side of things. I would do less on the landing page in regards to security and push that to more in depth detail in the docs. Talk about best practices of creating a read-only user. Create guides for most popular db vendors, etc
* As has already been mentioned, you will get concerns about any access to the db and "why not an API". I think these commenters are right that this will be a deal breaker for many companies, but I don't agree that you want an API. Not only because I think that reduces some of the value prop but also because then you would need to store customer data. I obviously don't know what you are storing today, but I would think that this model could have a big advantage if you didn't need to store any PII data in cc.dev. The complexity of dealing with compliance and regulatory requirements is no small part of why something like MailChimp is expensive is that burden. I don't know if this is something you currently consider a value prop (or if you have engineered to support this)... But I certainly would :)
* To address the reality of private dbs/giving access to a db, I would look at potentially implementing an agent. This agent would then run the queries (still provided by the user) and just-in-time deliver the results when needed. Getting the model of this right that works across different companies view of security will mean this is probably best to do with a handful of potential customers.
* As far as deployment and monetization model... I would only open source it if you are giving up on commercialization or if you can open source a subset that can be useful enough to build a community. Once again, returning to JAMstack, maybe solve in open source (but integrate) some problem unique to those teams. As far as a paid self-hosting, given that this seems like a one man show, I would resist it. Trying to do SaaS and support of self-hosting is rough, even for well funded VC backed teams..
Quite a bit here, but if you want follows up on anything, my email is in profile or will reply to comments.
Congrats again and hope it goes somewhere :)","created_at":"2023-07-23T18:03:32Z","created_at_i":1690135412,"objectID":"36837923","parent_id":36836596,"story_id":36836596,"story_title":"Show HN: I made a MailChimp alternative that connects to your database","story_url":"https://cc.dev","updated_at":"2024-09-20T14:37:05Z"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"the__alchemist"},"comment_text":{"matchLevel":"none","matchedWords":[],"value":"I will give this a try! I feel like AI should be able to automate this process (Schematic and/or PCB). I made a new design 2 months ago; tried to get Claude and COdex to do the schematic in KiCad, and they were not effective. Made a mess of it for an initial build, and edits later. I didn't bother with the component placement and routing.
I did find LLMs to be very effective on answering questions about routing, part selection, and identifying problems; was using a new MCU (Atmosic) with some unusual power circuitry, and got a working board first go. LLMs also crushed getting it working with rust (Very likely I am the first person to program an Atmosic MCU in rust); found a debugger that would work (NXP MCU-Link; the St-Linkv3s I normally use won't work for this, and the official recommended part is very expensive). The LLMs wrote a probe-RS configuration for it, linking scripts etc.
The LLMs also, under close direction, wrote a minimal HAL subset for I2C, SPI, power management, clocks, and BLE. I pointed it to my STM32 HAL as an API guide, set up the basic data structures, and pointed it to the Atmosic SDK (In C). It connected the dots, and produced a working HAL with a familiar API."},"story_title":{"matchLevel":"none","matchedWords":[],"value":"AI-assisted PCB design for KiCad 10"},"story_url":{"matchLevel":"none","matchedWords":[],"value":"https://github.com/mixelpixx/Konnect"}},"_tags":["comment","author_the__alchemist","story_50085553"],"author":"the__alchemist","comment_text":"I will give this a try! I feel like AI should be able to automate this process (Schematic and/or PCB). I made a new design 2 months ago; tried to get Claude and COdex to do the schematic in KiCad, and they were not effective. Made a mess of it for an initial build, and edits later. I didn't bother with the component placement and routing.
I did find LLMs to be very effective on answering questions about routing, part selection, and identifying problems; was using a new MCU (Atmosic) with some unusual power circuitry, and got a working board first go. LLMs also crushed getting it working with rust (Very likely I am the first person to program an Atmosic MCU in rust); found a debugger that would work (NXP MCU-Link; the St-Linkv3s I normally use won't work for this, and the official recommended part is very expensive). The LLMs wrote a probe-RS configuration for it, linking scripts etc.
The LLMs also, under close direction, wrote a minimal HAL subset for I2C, SPI, power management, clocks, and BLE. I pointed it to my STM32 HAL as an API guide, set up the basic data structures, and pointed it to the Atmosic SDK (In C). It connected the dots, and produced a working HAL with a familiar API.","created_at":"2026-10-16T13:40:59Z","created_at_i":1792158059,"objectID":"50118294","parent_id":50085553,"story_id":50085553,"story_title":"AI-assisted PCB design for KiCad 10","story_url":"https://github.com/mixelpixx/Konnect","updated_at":"2026-10-16T13:44:49Z"}],"hitsPerPage":20,"nbHits":26,"nbPages":2,"page":0,"params":"query=SEO+data+API+expensive&advancedSyntax=true&analyticsTags=backend","processingTimeMS":19,"processingTimingsMS":{"_request":{"queue":45,"roundTrip":15},"afterFetch":{"format":{"highlighting":6,"total":7},"merge":{"mergeLoop":{"prepareNextHit":2,"total":2},"total":2},"total":2},"fetch":{"query":8,"scanning":7,"total":16},"total":19},"query":"SEO data API expensive","serverTimeMS":72}