Ideas Lab

I left Figma to build a diffusion-based UI design tool

Hey hn, this has been something I've been working on for the last few months and is finally robust enough to really show off.I've been pretty tired with the design outputs of LLMs for a while, and I've always thought diffusion offered much more creative / on brand design outputs, even before they were able to render text.I had enough conviction for this to leave my role over at Figma to build Diffui. The goal is to allow for you to design your full web app as quickly as possible, in a figma-like interface, and then hand that off to an agent to build. The page shows some interactive examples. Happy to be an open book here and chat about the approach, the process, etc!

Hacker News · 3mo ago · Tool · AI

Trifle – Open-source analytics that stores answers, not events

Trifle is an open-source time-series analytics library that aggregates nested counters instead of storing raw events. All in the database you already have. After rebuilding it twice over 10 years, it now tracks ~1B events a day at my day job.It started in 2015 as my own Rails APM. I plugged into ActiveSupport::Notifications, got a few small users, and one bigger one whose scraping app broke everything. That sparked the core idea: aggregate counters into pre-defined time buckets, so a single write increments multiple buckets at once. The APM eventually faded away without much traction.Later in 2021 I needed analytics at my day job. Instead of going for something out there I revised the idea of Trifle as a more generic analytics library, borrowing some data warehouse ideas. First used Redis, then Postgres, eventually MongoDB. Hence why Trifle::Stats comes with multiple drivers that keep the DSL unified while storage layer changes with your needs. In our case (huge write volume, some reads) PG read faster but slowed on large writes.The nested values are the whole trick here. Single: Trifle::Stats.track( key: 'requests::aws::s3_uploads', values: { count: 1, status: { request.response_code => 1 }, size: payload.bytes, duration: { sum: request.duration, count: 1 } } ) builds up counts for requests, success rate, result status codes, duration for multiple time buckets at once. Single bucket from 2am then looks like: { count: 14, status: { 200: 12, 500: 2 }, size: 5628341, duration: { sum: 43, count: 14 } } If request.duration is in seconds, then sum stored under duration would be in seconds as well.Success rate is never stored, but it is calculated by dividing 200s over total number of requests. Same with average duration: sum over count. You ask for a metrics key, granularity and timeframe and you get back aggregated values at each point. Ready for charts or to answer "Average response time over last 30 days&qu

Hacker News · 3mo ago · Tool · AI

OneCLI – OSS credential gateway that keeps secrets out of AI agents

hey HN, Jonathan and Guy here, creators of OneCLI (https://onecli.sh/). OneCLI is an open source vault for AI Agents.Traditional vaults are used to store your secrets and, on demand, provide them to you all in a secure way, trusting the person to keep them safe. We figured that in the agent's world, this is not the case, as you don't know what happens with the secret after it's delivered to the agent, or where it was saved. Or maybe someone even manipulated them to hand them over...From that understanding, we decided to build a network gateway that sits between your AI agents and the services they call. OneCLI matches the request by host/path, verifies the agent should have access, swaps the placeholder for the real credential, and forwards the request. the secrets set inside the OneCLI vault, encrypted on rest, or could fetch in realtime from your bitwarden / 1password wallets.Demo - https://www.youtube.com/watch?v=5e5pbPEzZfY.We started working on this by accident, even though our careers were in the security space. We were working on a devtool called ChartDB, an open-source DB tool. When OpenClaw took off back in January, we started using it to orchestrate agents on top of ChartDB. We quickly understood there is a big issue around auth. Agents need credentials to do real work, but to give them those secrets would not be the best idea. they keep them in their memory and also write them down to local files and their sessions as plain text. And we knew that agents can easily be fooled into giving up those API keys/secrets. So we needed some way to control the agent and stop prompt injections from tricking it into using its services for an attacker's benefit. Not providing the keys to the agent + adding alerts or human-in-the-loop for sensitive operations, in case someone manipulates the agent, and access logs are being audited.We use it ourselves daily. My OpenClaw runs my day across Attio, Gmail, and

Hacker News · 3mo ago · Tool · AI

Echo – Fable-level results at 1/3 the cost using open-weight models

I’ve been building Echo (https://echo.tracerml.ai/), an experiment in making one AI system out of a pool of open-weight models rather than choosing a single model and using it for every task.It started with a simple experiment. I took a group of models, including GLM-5.2, Kimi K2.7 and others, and ran them on the same evaluations. Then I measured what would happen if, for each problem, you somehow knew in advance which models would be useful and how their outputs should be combined.That hypothetical system performed substantially better than any individual model in the pool. Of course, it is not something you can actually deploy because it relies on knowing which decisions were good after seeing the result. Echo is my attempt to recover some of that advantage without having that information in advance.For each request, Echo decides how much computation to allocate, which models should participate, and how their work should be combined. Some prompts may only need a relatively small amount of inference, while others benefit from multiple models working on different parts of the problem.One thing that surprised me while building it was how complementary the models are. A model that is clearly weaker overall can still be extremely useful on particular problems or as part of a combination.On my first evaluation mix, Echo consistently performed better than the best individual model in its pool. It also reached roughly the same aggregate result as Fable, which I used as one of the stronger comparison systems, at around one third of the inference cost.There are still some cases where Echo makes the wrong allocation or combination decision. I’m currently spending a lot of time understanding those failures, as well as testing whether the same approach holds up on coding and agentic tasks where measuring the quality of each decision becomes much harder.I built a chat interface (echo.tracerml.ai) and an OpenAI-compatible API (https://echo.tracerml.ai&#x

Hacker News · 3mo ago · Tool · AI

Is there a website that tracks excessive writes to SSDs in OS/app betas?

If not, why not?That way, people can choose whether they want to install various software betas based on how much SSD wear they are willing to accept.And since this data would be made public and available in one place, software developers would be more careful about excessive logging in their betas.

Hacker News · 3mo ago · Tool · AI

Palmier Pro – Open-source macOS video editor built for AI

Hi HN, we are Marcos and Harrison, cofounders of Palmier (https://palmier.io). We are building Palmier Pro, an open source macOS video editor, with built-in AI generation and a local MCP server that connects to your agent. Here are a few demos:- Making some AI transitions: https://www.youtube.com/watch?v=hbM_-eR1GX4- Multicam editing with Codex: https://www.youtube.com/watch?v=SjS2q2LT1q8- Cutting long form clips into shorts: https://www.youtube.com/watch?v=PR66eN2ouuQWe built Palmier Pro as an internal tool when we were making AI launch videos for other startups. The main problem it solved in the beginning was the back-and-forth between AI generation platform and video editor. The iteration loop was awkward: AI videos → download → import to editor → edit → realize we need to change the AI video → repeat. So we built a minimal video editor where we could let Claude generate AI videos inside the editor.As we gave more and more tools to the agent, we wanted to push to see what else agents can do in the video editing space. So today, your Claude/Codex can:- Manage projects inside Palmier Pro- Import media from a public URL or filesystem to the project, and organize them in folders- Search media (by embedding footages using SigLIP2 running locally)- Edit the timeline (tracks/clips/keyframes operations)- Generate images, videos, sound effects, captions, music- Export videosThere are two ways for LLMs to interact with the editor: by connecting to the local MCP server, or using the in-app chat. Both use the same tools and APIs exposed by the video editor.We have seen people using MCP server to connect to their own workflow to automate massive-scale video editing (e.g. given this same podcast style, replicate it with other footages that I have). We have also seen people using the in-app chat where it lives closer to the editor UI, with lower latency for faster iteration.We don't believe that AI is go

Hacker News · 3mo ago · Tool · AI

Screenpipe (YC S26) – Record how you work and turn that into agents

Hi Hacker News, I'm Louis. I built Screenpipe (https://screenpipe.com), an app that records your screen and audio locally (only!), and gives AI agents a searchable memory of what you've seen, said, and heard. This makes it easier to automate your repetitive tasks, turn them into SOPs (Standard Operating Procedure) and so on.I made a HN-style demo video at https://www.tella.tv/video/build-your-ai-second-brain-with-s... and there’s a marketing video at https://www.youtube.com/watch?v=c1jV6E9pyug.I’ve been obsessed with this for a long time. I’ve been maintaining a “second brain” since 2020, in which I would store journals, handwritten notes, music I listen to, projects I'm working on, conversations I have with people, personal CRM etc. I experimented a lot of RAG in the early days with ParlAI, hundreds of fine-tuned GPT2 models, and GPT3 (https://forum.obsidian.md/t/fine-tuning-openai-api-gpt3-on-y...). Later I built Ava, the first Obsidian AI plugin, which grew to a few thousands of users quickly. It then became Embedbase, an API to make it easier to build AI apps powered by RAG.What I learned from all this is how important it is for the models to have context about what you’re doing on your computer, in order to get them to do what you want.In the early days there was fine tuning but it was too much pain, then there was tool calling so that AI can access software you use but still kinda not autonomous enough. needing micro management. Then MCP came, but it felt too static, and non technical users struggled to build and use MCP. Then we got skills. Most recently we’ve seen Karpathy’s LLM-maintained wiki, Garry's GBrain, etc., where an agent incrementally maintains a persistent collection of Markdown pages. New sources update entity pages, strengthen or contradict existing claims, and improve a synthesis that compounds over time. I like this pattern, but it still begins with someone s

Hacker News · 3mo ago · Tool · AI

Cactus Hybrid: We taught Gemma 4 to know when it's wrong

Hey HN, Henry & Roman here from Cactus.A small, on-device model is fast and private, but sometimes wrong, but frontier models are getting expensive pretty fast. So, we post-trained Gemma 4 E2B post-trained to know when it's wrong. Every response comes with a confidence score between 0 and 1. Developers can accept the on-device when it's high, hand off to a bigger cloud model when it's low. By routing only 15-35% of queries to Gemini 3.1 Flash-Lite, Gemma-4-E2B matches Gemini 3.1 Flash-Lite on most benchmarks.- ChartQA: 15-20%- LibriSpeech: 25-30%- MMBench, GigaSpeech, MMAU: 30-35%- MMLU-Pro: 45-55%We were always frustrated by the routing signals hybrid apps rely on: asking the model to rate itself in text (unreliable, and you're parsing prose), or token entropy heuristics (barely better than a coin flip in our tests). So we did mechanistic studies on small models, Gemma 4 particularly, and found the hidden state for different layers carry meaningful self-awareness signal for various situations.SO we extended the model with a 68k params probe layer (LayerNorm, low-rank projection, attention pooling, small MLP head) reads one intermediate layer during decoding and predicts p(wrong); confidence = 1 - p(wrong), returned as structured data, never parsed out of the answer text.Across 12 hold-out benchmarks spanning text, vision and audio, the probe averages 0.814 AUROC vs 0.549 for token entropy. The result that convinced us this is real: the probe was trained on zero audio data, yet scores 0.79-0.88 AUROC on four audio benchmarks where entropy is near-random or worse (0.32-0.52). It's reading a modality-independent correctness signal from the hidden state, not memorizing patterns from its training data.We published all weights on HuggingFace and provide copy-pase codes to run it on Transformers, MLX, Llama.cpp or Cactus. With Ollama, vLLM, SGLang etc in the works. For llama.cpp we ship a patch series you compile in once (upstreaming is planne

Hacker News · 3mo ago · Tool · AI

Word in Web – Near MS Word Parity Docx Editor in Web

Word in web is a pure JS docx editor/renderer benched against MS Word directly. This was largely inspired by Eigenpal going closed source, and some personal frustrations I had with working on complex Word templates, pleading papers, etc... and not having a clean way to view/edit them without owning MS Word (which I did just eventually buy but it sucked).This was also an exercise in highlighting the value of good evals for Agents to bench against. Instead of just throwing the OOXML standard at an agent and letting it spin, I put together a large corpus of Microsoft Word docs - converted them to pdf and then rasterized them to images. The eval then compared the image diffs measuring the actual differences in the actual pixels. This let me (and more importantly, the agent) catch a lot of the edge cases and optimize aggressively while preventing 'whack a mole' style regressions that would have made this project a lot harder.Edit: I forgot to mention two things, 1. I published the full eval, including the raw diffs and delta over time so you can see all of the fixtures I actually tested against.2. On the DX side, it ships as a typed React package and runs entirely in the browser. You can use it as a viewer or editor, customize the styling and visible toolbar features, build your own controls against the API, and decide where the edited DOCX bytes are saved, so it’s relatively easy to drop into another project

Hacker News · 3mo ago · Tool · AI

Bento - An entire PowerPoint in one HTML file (edit+view+data+collab)

Over the past few months, our team has been building more and more slidedecks using web frontend technologies with coding harnesses like Claude Code, but a common complaint is to make even small edits we need to edit the code either manually or via the harness.To avoid this loop, I ended up creating Bento, a single HTML file with everything you need in a slide tool including animations and shared editing. There's no install or cloud login, everything works offline. The default deck is around 560 KB and it doesn't need to fetch anything once you got it.Open it in a browser and then you can edit, present, print and save. Share it via email or via Airdrop and all they need is a browser to edit, present and also do live collab on the slides. Drop it in to Claude or ChatGPT to transform existing pptx files into Bento slides. There is no cloud involved, only an encrypted blind relay to allow for shared editing. The relay doesn't see any of the data.Check it out at https://bento.page/slides/ which takes you straight to the editor.Go to https://bento.page/guestbook/ to try out the live guestbook to experience share editing / collab.There is also a gallery with some sample decks on the website - https://bento.page/All the code is MIT licensed and you can find it here - https://github.com/nyblnet/bento . I used reveal.js with several other libraries (including some homegrown ones), and Claude Code.

Hacker News · 3mo ago · Tool · AI

Q3Edit – Edit and play Quake 3 maps in the browser

I've been building a level editor for Quake 3 that runs entirely in the browser: Radiant-style layout, brush and patch editing, CSG, terrain sculpting, and entity editing. It opens and saves .map file and you can play the maps you build directly in the browser using a webassembly build of ioquake3.

Hacker News · 3mo ago · Tool · AI

A new kind of FPS aim trainer

I played a lot of Valorant and got mad, so I made an aim trainer that analyzes your raw crosshair movement to explore your raw motor and perceptual weaknesses instead of scoring scenarios. It also chooses sens and difficulty as part of the tasks, and makes playlists that are optimal difficulty for you to learn and progress faster!

Hacker News · 3mo ago · Tool · AI

DeepSQL – A self-hostable DBA agent for Postgres and MySQL

Hi HN - I'm Venkat, founder of Stayflexi (YC), CMU CS grad and Ex-Oracle Query Engine team (patents in core databases)DeepSQL started as an internal tool to stop our own databases from becoming the bottleneck they were becoming (13,000+ hotels in production). It worked well enough that we're releasing it.DeepSQL is an AI agent that operates a database the way a senior DBA and Data Engineer would1. Fixes slow queries (we cutdown DB spend by 4x) 2. Fixes DB bloat (blocks unnecessary schema changes, in vibecoded setup) 2. BI dashboards(we removed spend on tableau, retool and appsmith) 3. Security (We could redact access to sensitive, PII data to our employees) while engaging everyone in the org to interact with our databaseHow it works 1. Deepsql learns your data and relationships from your code base, rules and query logs 2. Deepsql agent has 20 background jobs that continuously monitor the schema changes, bottlenecks, and come up with solutions before the problem arises. (Schema bloat issues are irreversible) 3. Deepsql Brain has CLI and MCP surfaces that can directly work with Claude / Codex and Cursor. 4. Deepsql digest delivers the daily database health report. 5. Your team can connect to deepsql with their favourate surface areas - Web UI, CLI, MCP or Slack.One line install: curl -fsSL https://install.deepsql.ai/install.sh | bashFor community support: https://discord.gg/duEJq7AeeGOr just email me for expert level setup guidance venkat@deepsql.ai

Hacker News · 3mo ago · Tool · AI

Unlayer (YC W22) – Add email and document builders to your app

Hi HN, We’re Adeel and Umair, co-founders of Unlayer (https://unlayer.com/). We let you add content creation to your applications without having to build an entire editor, renderer, template, and export stack yourself. Unlayer lets you create emails, web pages, and documents inside your app, in three different ways: in code, visually, or with AI.Here’s a demo: https://www.youtube.com/watch?v=0HsDtNkdMpM.We started with an embeddable email editor because a lot of products eventually need one: CRMs, marketing tools, customer engagement platforms, marketplaces, internal tools, and vertical SaaS apps all run into this at some point. At first, it sounds like a small feature: "just" add a drag and drop editor. In practice, it turns into a big pain. You end up dealing with email rendering, Outlook quirks, responsive layouts, templates, merge tags, image uploads, exports, permissions, localization, versioning, and a long tail of edge cases that have nothing to do with your core productOver time, we saw the same problem beyond email. Apps also need landing pages, invoices, proposals, reports, contracts, and PDFs. Some of this content is best created visually by end users. Some of it is better generated in code by developers. Increasingly, some of it is also generated by AI agents. Many teams eventually need all three workflows. That is the direction we have been working toward with Unlayer.There are three parts we are showing today:(1) Unlayer Elements. This is our open-source React component library for creating emails, pages, and documents in code (repo: https://github.com/unlayer/elements, more at https://unlayer.com/elements). Instead of hand-writing raw HTML templates, developers can compose content using React components, reuse sections like headers, footers, CTAs, invoice rows, and branded blocks, keep templates in Git, and render them into production output.One newer use case we are seeing is

Hacker News · 3mo ago · Tool · AI

CodeAlmanac – Karpathy-style codebase wiki from your conversations

Hey HN! This is Divit from Almanac (YC S26). We built CodeAlmanac, a wiki for your coding agents that updates as you talk to them. It is open-source, local, and free.Here’s a demo: https://www.youtube.com/watch?v=XNQWV3TFBWMYour CC/Codex conversations contain a LOT of knowledge that is forgotten because it was never documented. People have their own methods of documenting their chats. We used to make Markdown files like MANUAL.md and DESIGN.md, and would prompt Claude to keep them updated. The problem is that these files quickly become outdated and messy, and there’s only so much you can put in a single file.So we set out to build CodeAlmanac. We wanted something that was 1) maintained automatically, 2) lived inside our repository, and 3) used our existing Codex/Claude Code subscriptions.CodeAlmanac maintains an almanac/ folder inside your repository. It contains connected Markdown pages that cover things not documented in the codebase, including decisions you have made and why the codebase is shaped this way.The pages are indexed in SQLite and are queryable through a CLI. We add instructions to AGENTS.md or CLAUDE.md so future sessions automatically search the wiki before they start coding.Every five hours, CodeAlmanac uses the Codex/CC SDK to spin up an agent that reads your new conversations and updates the relevant pages. We went with a time-based trigger instead of commits because we saw people commit very frequently, which would lead to high token costs.We originally made CodeAlmanac for individual developers, but the team use case has become much more obvious to us while using it ourselves. We are a team of three, and each of us works with our own coding agents. Before this, I would make some change after a lot of thinking and then later have to call my cofounders and explain why it looked this way. There is almost a sense of relief now knowing that the decisions I made are written down somewhere their agents will actually r

Hacker News · 3mo ago · Tool · AI

Bloomy (YC S26) – AI-powered mastery learning for K-12

Hi HN, I’m Alex Southmayd, the founder of Bloomy (https://bloomylearning.com) – an AI-powered mastery-learning platform for K-12 students. Bloomy provides students with an AI tutor alongside adaptive curriculum (right now Math, English Language Arts, and Writing).How it works: we diagnose students’ skill gaps, place them on personalized learning paths, and give them standards-aligned lessons and a Socratic AI tutor that scaffolds their learning without just giving away the answer.The goal is to solve the Bloom 2-sigma problem (https://en.wikipedia.org/wiki/Bloom%27s_2_sigma_problem) with AI.Short launch video: https://tinyurl.com/bloomylearningLonger product demo: https://youtu.be/XHvoKt6qMeoFamilies access for Bloomy: https://bloomylearning.com/familiesI started as a teacher. I taught 7th-grade English and writing with Teach For America, and every day I struggled to deliver differentiated instruction to 30 students with 30 different sets of needs. Some students needed remediation, some needed acceleration, and many needed a tutor sitting next to them helping them reason through the next step. Benjamin Bloom’s two-sigma result—that one-on-one tutoring can produce much better outcomes than conventional classroom instruction—always felt intuitively true to me. The hard part was making that kind of attention affordable and available to every child.Then AI changed the cost curve. When I saw schools such as Alpha organize academics around mastery rather than seat time, the model clicked. If you’ve heard of Alpha School, that is directionally the kind of learning model that inspired us. But I kept thinking about the families and schools that already exist: homeschool families, microschools, hybrid schools, and regular classrooms where most children are today.Most students and teachers see learning gaps at the wrong resolution. They get a grade, percentile, benchmark score, or broad standard—not “th

Hacker News · 3mo ago · Tool · AI

How do people keep track of organizational knowledge?

With code coming in faster and faster, we've been losing track of context, what's out of date, the source of truth, etc. Also with individuals spending less and less time on any one problem, finding an expert to answer questions has become challenging. People don't have that single domain of expertise like we used to. There's too much surface area to cover.Are we the only ones dealing with this? How's everyone else handling it?

Hacker News · 3mo ago · Tool · AI

Justif – Knuth-Plass justification and microtypography for the web

Justif is a drop-in JavaScript library that progressively enhances web pages to TeX-level text justification. Installation is a single <script> line, standard text and accessibility affordances are unchanged, and users with JS disabled get native browser rendering.I made justif because I've long been a fan of justified text. I think it looks clean and elegant, and makes reading more enjoyable. But bad justification is the opposite, with gaping spaces that distract me to the point of making the text near unreadable.Browsers have got better recently at handling justified text [0][1], but still use suboptimal greedy algorithms for the most part, and are not near that TeX/InDesign level of quality that I crave. `text-wrap: pretty` exists but is far from a panacea, as you can see for yourself in the demo.Justif also has the benefit of providing more consistent text layout across browsers. Blink (Chrome), Gecko (Firefox) and WebKit (Safari) all justify text differently, so normally what the user sees depends on what browser they use.Take a look at the demo, play around with it, see if you can break anything. I'm open to improvements of the API design as well, so please let me know your thoughts.For more details, see the README: https://github.com/lyallcooper/justif0: https://cloudfour.com/thinks/justified-text-better-than-expe...1: https://owickstrom.github.io/the-proportional-web/

Hacker News · 3mo ago · Tool · AI

LoopGain – Stop agent loops with control theory, not max_iterations

Hacker News · 3mo ago · Tool · AI

I replaced a $120k bowling center system with $1,600 in ESP32s

I might be the only SRE on Earth with his own bowling center. It's a more in-depth gig than you'd think.My family and I bought an abandoned 8-lane bowling center in the rural mid-west. In our small town there weren't many recreation options for families. You've heard of a food desert? This is an R&R desert.It had been abandoned for a good reason. The roof leaks, the electrical system was constantly surging, and my 70-year-old bowling equipment (still) doesn't work perfectly. The system that keeps your score is particularly interesting to me. It's the thing you watch during your game, but it fades into the background beyond that. Turns out these things are really cool, but absurdly expensive.Ours was installed in 2008 and cost six figures. It's calculating ball speed and trajectory, camera-based pin detection (object detection and trig, on ICs!), runs the fouling, the animations, the pinsetting machine and ball return. Very cool stuff for its age.From the business perspective, my facility only cost me $105k. To forklift-replace the score keeping system runs anywhere between $80-$120k, depending on features, vendor, and unit age. No upgrades or service contracts, mind you, and every feature and customization is a new line item. That's for a 1:1 replacement on a system installed in 2008. Incredible, given how fast the tech world moves.Replacement parts cost a shocking $4000 per pair of lanes. But wait, the bowling machines themselves are 70 years old, so what's this "advanced" system actually doing back there? Actuating a single relay to trigger that big old machine. Everything else is strictly mechanical. Meanwhile I've got a six-figure invoice in my hand. I'm upset.Given the state of open hardware, computer vision, real-time event streaming, and open source running megascale products worldwide, there had to be a way to do this myself.So far I've built an equivalent prototype for about $200 per

Hacker News · 3mo ago · Tool · AI