Live data from Hacker News

Ask HN: What are you working on? (September 2026)

news.ycombinator.com

641–650 of 870 posts

Re: Ask HN: What are you working on? (September 2026)

#641

I'm buildingan open-source way to fill U.S. immigration forms without fighting clunky PDFs Github: https://github.com/athos2113/fillvisa-os Demo: https://fillvisa.com/demo/ US Immigration still relies on outdated XFA PDFs - you can't fill them on your browser. Most immigrants end up printing the form and fill it manually. So, I converted the PDF forms into smart web forms. They are replica of the official USCIS forms…

Apple preview and the add text button got me through all the USCIS forms. You can use AI computer control to do this stuff now, right?

> You can use AI computer control to do this stuff now, right?

Speaking as someone who had to do a lot of immigration work to move to Japan, I can think of no work I would want to have a close eye on than immigration paperwork. One error is all it takes for a clerical rejection or similar. I had a professional lawyer working on mine and still cross checked every detail, and found errors!

Re: Ask HN: What are you working on? (September 2026)

#643
post #612

https://shopspec.io/ ShopSpec is a deterministic parametric furniture generator (bookshelves, cabinets, shop carts, leg-and-rail frames, multi-carcass composites) that encodes real woodworking intelligence and outputs shop-ready artifacts. Most recently, introduced the concept of value engineering by sweeping across dimensions to batch evaluate candidates that ranks based on sheet purchases. Also have in beta an MCP…

Wow that's really cool! I'm actually trying to learn about woodworking and using those example as a base are really great! You could consider having an option for metric users as well!

+1 for metric. This looks really cool, indeed.

Re: Ask HN: What are you working on? (September 2026)

#646
https://www.cellaflow.com/

Cellaflow — Safe retries for AI agent tool calls.

The question I'm interested in is: how do you make sure recovery of an AI agent doesn't repeat the side effect?

Recovering an agent's state after a crash is the easy half, and most frameworks already have a checkpointer for it. The hard half is recovering after the agent has already done something irreversible.

Sometimes the downstream API has idempotency built in. But often you don't control the API, the operation isn't naturally idempotent, or you need to coordinate several side effects. Sending an email, creating a ticket, triggering a webhook, provisioning something, or mutating another system can all leave you with the same problem.

Cellaflow focuses on the side effect, rather than just the agent's state: which agent is allowed to perform an operation, what's already been done, and how to safely recover when the agent crashes or disappears mid-execution.

It is a runtime that sits underneath your existing agent framework, so you can use it with LangGraph, CrewAI, AutoGen, or your own agent code without migrating the agent itself.

Still early. I'd like to talk to anyone running agents in production who's hit duplicate tool calls, duplicate side effects, or a workflow that quietly never finished.

Re: Ask HN: What are you working on? (September 2026)

#647
https://storyofus.app

I saw one Anthropic MTS got married recently and he did bunch of relationship analysis through iMessage dump in Claude Code. Started with that idea, but I don't want AI to have all my personal chats, so gave it a local first discovery and just statistical AI and I really like it right now.

Re: Ask HN: What are you working on? (September 2026)

#648
post #63

Recently got genomic DNA purification automated for yeast (plus have automated library prep for nanopore+ a promethion)! As part of the DNA purification task, I had to journey up to Healdsburg to buy some lysis enzyme (thank god for winemakers, 100x cheaper than the biotech grade stuff), and noticed catalogs of different yeast strains with all their flavor profiles and stuff characterized. I'm thinking about buying a…

Super cool

Re: Ask HN: What are you working on? (September 2026)

#649
post #521

Earlier quoted context omitted.

I presume you mean SEO (Search Engine Optimisation) rather than SSO (Single Sign On). Six years ago I launched a boutique search engine which downranks pages which contain adverts on them (the theory being that this would remove the incentive for creating spam) and of course contains no adverts itself (to avoid what Brin & Page described as being "inherently biased towards the advertisers and away from the needs of c…

Yes, SEO, wrote the comment in rush, sorry. I am running a very popular dance calendar site in Finland that contains very relevant information, in very condensed format, but has some ads that are highly relevant for readers. I see why you might want to avoid ad based sites, but I think there may be some relevant sites that you will miss. The site https://tanssi.io

Pages with adverts are downranked (i.e. likely to appear lower in the results) rather than hidden, so if there are enough other positive signals they could still appear higher in the results.

Re: Ask HN: What are you working on? (September 2026)

#650

Earlier quoted context omitted.

But how would you know if someone was just lying when they added "[HUMAN-CODE]"?

People don't say when they vibe code things but it's still very obvious.

Vibe code, maybe.

But a lot of developers use AI as a pair programmer, and I doubt you'd be able to tell the difference.

Post reply on HN