Ask HN: What are you working on? (April 2026) (Non AI)
Tell me about your latest non-AI project and I'll check it out
It's a free USCIS form-filling web-app(no Adobe required). USCIS forms still use XFA PDFs, which don’t let you edit in most browsers. Even with Adobe, fields break, and getting the signature is hard.
So I converted the PDF form into modern, browser-friendly web forms - and kept every field 1:1 with the original. You fill the form, submit it, and get the official USCIS PDF filled.
I found out SimpleCitizen(YC S16) offers a DIY plan for $529 [2]
So, a free (and local-only) version might be a good alternative
I used a unique modern matrix and polished everything with another project I built which's a deterministic picture generator used to let people interact with something interesting, already using it to produce their default profile picture using their id as the seed.
It meant to run on raspberry pi and I already built an img for the first release.
https://github.com/remohexa/rematrix-gallery
And there's also a demo: https://rematrix.remohexa.com/
I actually like your remohexa.com website how it seems to change colors and the dropping emojis and colors are appealing. I did get the fuschia version at first and it was eye-catching I liked it a lot. I got the green version a second time and was like oh what is the dev actively changing it right now. Then I realized keep hitting refresh for different colors.
I genuinely don't really know how to make videos that might catch the viewer eye, and overall I don't really have a "good enough" equipment or a place to make videos. But I feel like your suggestion for yt was the missing piece that I didn't think of. Which I'll definitely take into consideration.
Thanks for your feedback about remohexa.com too, I'm glad you liked it. The changing colors idea was just a totally random thought while I was making that page, but it might be also a good idea to implement some kind of accent color that changes with each visit (not each refresh) for the real website too. Don't you think?
A lot of tools like Lovable and Bolt generate apps that look great but don’t expose content well to search or AI crawlers. The impact is lower search ranking, fewer AI citations, and broken social links.
We fix all of that by returning fully rendered HTML to SEO bots and clean Markdown to AI crawlers so they can actually understand the content. We also fix the Social Sharing broken links with custom HTML generated just for sharing your links on social.
Quick setup, no code required.
Free: visibility test: https://datajelly.com/#visibility-test Our Edge Product: https://datajelly.com/products/edge
Happy to answer questions.
Built it because I was tired of the "just password protect it" workaround when sharing early stage work. Passwords get forwarded, and wiring up auth on something you'll throw away is wasted effort. I wanted a paper trail. Who saw what, when they signed, and a way to revoke access instantly.
Free tier gets you 1 project, 5 invites, default NDA template, and a 7 day audit log.
I've just launched it on iOS and Android in Germany with support for 10 languages. Happy to receive any feedback
We use data models and release rules borrowed from k-anonymity techniques, batched releases and privacy pass cryptographic tokens to create super safe surveys, and everyone who participates as an invite peer gains access to the same full-fat report.
Our form supports specific benefits extension by geography, an extended equity compensation set of questions for packages where equity is significant; and performance pay questions for groups (like sales, execs) where performance pay is also a significant part of the package
Also, we make it easy to explore pay gaps (gender, ethnicity, gender identity, whatever it is) because you can run several peer groups under one poll. no person gets tagged but you do know which respondent was in which peer group - so you can keep context in view (but size needs to be at least 4) but also have a broader view by rolling up results.
Knowledge is power, and all that
We’ve spent a lot of time squeezing every bit of efficiency out of the hardware. By shifting the heavy lifting from memory-hungry string parsing to hardware-accelerated bitwise math, we can scan 1TB of data in about 1.2s for roughly $0.01. It’s been a fun challenge to see how far we can push the physics of the network and CPU to make searching massive amounts of raw data feel instantaneous.
If you’re dealing with similar scaling headaches and want to chat about it, my email is dhruv [at] roverhq.io or you can find more at https://roverhq.io/.
One feature that turned out to be particularly useful is what I think of as “blast radius”, If you change a function or component, what other parts of the system are affected?
This has been especially helpful when thinking about PR reviews, not just what changed, but what the change might impact indirectly.
It’s still early, and I am working on the cloud version which will include LLM interface to interact with graph and other team collaboration feature and github connection but I’d really appreciate feedback from others who might find this tool useful, and I invite anyone and everyone to contribute to the open source github repo.
Github: https://github.com/devlensio/devlensOSS Landing page: https://devlens.io/
I started it because existing writing tools didn't handle Japanese-specific needs well. It's fully local-first no account or internet needed.
The problem: sub-$2M ARR founders track competitors in a Notion doc that is always 3 weeks out of date. Everyone knows it is broken. Nobody fixes it until a deal falls through.
Peerscope monitors competitor pricing pages, feature pages, and messaging - sends plain-English alerts when something actually changes. No dashboard to check. No data dumps. Just what you need to act on.
Still validating demand before building. Waitlist at https://peerscope-waitlist.pages.dev
1) Etch-Detect API changes automatically using real traffic (no tests required)
https://github.com/ojuschugh1/etch
2) GhostDep-Detect phantom and unused dependencies across multiple languages using a fast Rust-based CLI
https://github.com/ojuschugh1/ghostdep
3) a local CLI that verifies whether AI coding agents actually did what they claimed
I'm building this site to display behavioral therapy data better for parents / caregivers. The site itself is just Next.js but the essence is ETL and data visualizations using the Observable Framework[1]. There is an opportunity to apply AI for Q&A.
Check out the newsletter for weekly updates!
More recently, I've made it so you can drop your existing playwright test suites into the code editor, and it'll Just Work.
A whole bunch more work to do around that, but I think letting folks drop code in makes more sense than continuously updating the UI.
The main thing I’m trying to improve right now is speed: fewer steps, better defaults, and outputs that don’t feel templated.
I'm still working on it. bulk create quote image featrue will coming soon.
Manage Cloudflare, Anywhere
https://apps.apple.com/us/app/cloudflare-remote/id6743181258
https://play.google.com/store/apps/details?id=com.emrekeles....
I am currently rewriting+testing the engine and about to add ~400 games to my platform in a few weeks.
A newsfeed for Substack posts from the past 24h. Its helping me discover writers other than just what the algorithm gives me.
Feedback board or feature board are the terms, and there's a plethora of SaaS options, from tiny to large.
I recently searched for one, but they were either over budget, required extra user accounts, weren't GDPR compliant, or too complex. In the end, I coded a custom solution for my site.
Yours would have fit if I had found it earlier.
I liked nolt.io from the design and stuff. But it does too much, and the monthly price didn't fit for a simple test. The others in that league were the same. The indehacker's are mostly not GDPR-compliant and have this doom-laden smell of neglect.
I was tired of copying and pasting SVGs over and over again for my projects and social links. So I built that. Tell me if you need any other SVGs.
Since last month we finally got our production API Key for EUSP/STAAN (it was certainly the slowest and most complicated search provider to adopt, so far), and that brought us to 5 search providers you can choose from and sort as you prefer.
We already have got over 40 paying customers (excluding family and friends, we’re guessing these paying customers came from some privacy listings and HN comments) and have exited beta last month!
Customers seem to really enjoy the simple UI (search can be used without JS) and search personalization (from choosing the providers to the domain boosting and exclusion). We also have hashbangs (like "!g", "!d", or “!e”) when something doesn’t quite give you what you’d expect, though.
You can see the main differences between Kagi and Uruky in the linked page, but one huge difference is that with Uruky, after being a paying customer for 12 months, you get a copy of the source code!
One thing we’re struggling with is outreach because we want to do it ethically, and it’s hard to find communities or places to sponsor which are privacy-focused and don’t require €5k+ deals. Ideas are welcome!
Because of bots there isn’t a free trial easily available, but if you’re a human and you’d like to try it for a couple of days for free, reach out with your account number and we’ll set that up!
Thanks.
P.S.: Because people have asked before, our tech stack is intentionally very "boring" (as in, it generates and serves the HTML + bits of JS to enhance settings and such), using Deno in the backend (for easier TypeScript), PostgreSQL for the DB, and Docker for easier deploying.
P.P.S.: Because this has been also brought up before, the name has no special meaning but we read it like "Euro-key" in English. Names are hard, and we’re aware it can remind people of Uruk and Uruk-hai. That’s OK.
P.P.P.S.: Another frequent question here is “how does it work?” When you search, we query the first search provider on your list, and if it yields less than X results (only Mojeek really gives us a total count, we have to try + estimate for the others), we try the second, and so on. We then merge the results in a round-robin fashion (first of first, first of second, second of first, second of second, and so on). There’s a bit of more nuanced logic to also properly rank the results with the pin/exclude/raise/lower preferences, because it works differently across providers and not all of them support that, for example.
Btw. Nice to meet you.
So what did you learn in SEO? I generally don't have much knowledge about it.
I would love to use it, it will certainly help me when I am not able to use my laptop.