Ovi Bodea

▪️maker-of-things▪️seeker▪️ reader▪️dog-lover▪️ traveller▪️ learner▪️teacher

[Fernando Rojo @fernandorojo]

We just launched 𝚟𝚎𝚛𝚌𝚎𝚕-𝚌𝚘𝚖𝚙𝚘𝚜𝚒𝚝𝚒𝚘𝚗-𝚙𝚊𝚝𝚝𝚎𝚛𝚗𝚜: every lesson from the talk below, now available as a skill.

Turn your React code into something you (and your LLM) enjoy working with.

▲ ~/ npx skills add vercel-labs/agent-skills

[Min Choi @minchoi]

Google just dropped Gemma 4 12B.

This AI multimodal model runs locally on your laptop without heavy encoder stack.

Vision. Audio. Reasoning. Agents.

Apache 2.0 open source.

[Markus J. Buehler @ProfBuehlerMIT]

We've made a breakthrough in self-evolving AI scientists moving from "search" to "principled discovery": Scientific discovery requires that the search space itself changes, and an AI scientist must perceive this shift without intervention. We built an AI that achieves this for the first time with the ability to discover the scientific vocabulary it reasons in. Evidence, tools, artifacts, verifiers, failures & claims become typed provenance. We show three distinct modalities: 1) retrieval, adding known objects; 2) search, exploring a fixed schema; and critically: 3) discovery, a verified regime transition.

We solve the open-endedness evaluation problem by lifting agentic workflows into a typed copresheaf and proving, via a Kan obstruction, that true discovery is not unbounded generation but a verifiable schema expansion: old evidence is transported by Left Kan extension, and genuine novelty is mathematically quantified by the pointwise residual beyond the transported image - separating discovery from mere search and making novelty objective and measurable rather than a subjective judgment or benchmark delta.

Our AI scientist is built in a way that does not pre-conceive the approach it chooses; instead, we endow the system with formal power to adapt, evolve, and reason from first principles. Case studies include:
1⃣Builder/Breaker model that discovers mode-conditioned compliance in proteins;
2⃣CategoryScienceClaw that finds anisotropic fiber-network stiffness rules.

Great work in collaboration with my graduate student @fwang108_ @MITdeptofBE

F.Y. Wang & M.J. Buehler, Self-Revising Discovery Systems for Science: A Categorical Framework for Agentic Artificial Intelligence, arXiv:2606.01444, 2026

[Dan Levine @daniel_levine]

I trust this infra team

[Margelo @margelo_com]

Ever wondered why animations on #reactnative's new architecture were so laggy?
We partnered with the @discord team, found the bottlenecks, and helped make their mobile app feel buttery smooth.
Here’s the before/after 👇

[Tuomas Artman @artman]

Today we are launching the inaugural version of Linear Diffs, our take on code reviews. Diffs is meant to enable product teams to accelerate shipping by making code reviews fast, focused, and in context.

Code review has remained a painful bottleneck while the rest of building software sped up. Growing volumes of code from agents are making it worse.

We designed Diffs around what a code review should actually be: instant to open, stripped of noise, and deeply connected to the issue, project, and customer signal behind the change.

It brings reviews inside Linear with smart prioritization, guided chapters for large diffs (following the logic of the work), structural highlighting that removes formatting churn, and rich context.

Agents already handle most of the line-by-line correctness. This gives reviewers the space to focus on the judgment that actually matters: architecture, fit, and real customer problems.

Available on all plans today. More of the workflow to come.

[Mitchell Hashimoto @mitchellh]

I've got an agent in a loop optimizing a renderer with the goal to minimize frame times (and tests to measure). It got times down from 88ms to 2ms and allocations down from ~150K to 500. Sounds good, right? Wrong. This is exactly why agent psychosis is a big fucking problem.

As an experiment, I rewrote the Ghostty core render state in Go, with access to identically laid out data structures as Ghostty and the exact same validation tests. I made a purposely naive renderer (simple, correct, but slow). 88ms per frame with 150,000 allocations (horrendous, lol)!

I then kickstarted a Ralph loop to bring the frame times down. I told it it can't modify input data structures or the public API or tests (they're correct), but it can do anything else it wants. It got to work.

It has worked for about 4 hours. I've spent around $350 on this experiment so far. The results?

88ms => 1.5ms
150K allocs => ~500 allocs

Incredible right? Nope.

My hand-written renderer I ported has frame times (same benchmark) of ~20us (0.020ms) and 0 allocations in the update path.

This is the problem with psychosis and lacking systems understanding. If you don't understand the system, you're going to accept that this is an incredible result. If you understand the system, you'll see better solutions immediately and can do roughly 75x better on throughput.

The people who blindly trust agent output are in the former camp. They're sheeple, overdrinking from a fountain of mediocrity.

Standard disclaimer: I use AI all the time. I like AI. The point I'm making is to not blindly accept results. Think. Analyze. Learn.

[ClaudeDevs @ClaudeDevs]

New in Claude Code (research preview): dynamic workflows.

Claude writes an orchestration script on the fly, then spins up a large fleet of coordinated subagents in parallel to take on your most complex tasks.

Use the word "workflow" in a prompt to get started.

[Jay Meistrich @jmeistrich]

Legend List 3.0 is here! 🎉

✨ React DOM support
✨ Even faster, more stable
✨ Perfect initial scroll
✨ KeyboardAwareLegendList for chat and AI apps
✨ Reanimated item transitions
✨ So much more!

It’s been almost 1,000 commits since v2, super excited to finally release it!

[Jay Meistrich @jmeistrich]

Finally! After years of searching for the best way to build the fastest mac markdown editor, I found it:

React Native.

30 MB app, opens a 10 MB file in 20 ms, while most other apps freeze or crash. And no webviews.

What do you want in a Markdown editor besides being fast AF?

[Marc @mrousavy]

The fastest markdown editor, powered by Nitro Modules! 🔥

[Software Mansion @swmansion]

react-native-streamdown v0.2.0 is out! 🚀
Powered by the new react-native-enriched-markdown v0.6.0, markdown now streams more smoothly with the full GitHub Flavored Markdown (GFM) support, including seamless real-time rendering for tables and rich formatting.

If you’re building LLM or AI-chat experiences in React Native, this update is for you. Check it out! 👇

[Ovi Bodea @ovidbme]

now this is interesting. would have probably been way more successful if it was the other way around. i’m surprised that AWS would open source something that could allow people to migrate from DynamoDB to something else

[AWS Open Source @AWSOpen]

🚀 Just launched: ExtendDB — an open source DynamoDB-compatible adapter written in Rust.

✅ Full wire-protocol compatibility ✅ PostgreSQL storage backend ✅ Pluggable architecture for more backends ✅ Works with existing AWS SDKs & CLI

Apache 2.0 | v0.1 — come build with us 🛠️

https://t.co/U6xouvSRwX

[Ovi Bodea @ovidbme]

@ty_todd1 interesting, i’ve been using GEPA for building plans. can you give some insight into how you’re using it to explore datasets?

[Ovi Bodea @ovidbme]

@steipete @ycombinator @conductor_build btw the main reason i like it so much is because worktrees are the front and center, and switching between them is one click. super easy to work on multiple projects and on multiple features in the same project.

[Ovi Bodea @ovidbme]

@steipete @ycombinator have you tried @conductor_build ? i used cmux before it, but now i dont think i want to use anything else for coding

[Ovi Bodea @ovidbme]

@Bencera hi neb! i’m impressed

[Ovi Bodea @ovidbme]

one founder 30M raised and the company name.. check this out AI Slop spelled backwards 🤦‍♂️

[Ben Cera @Bencera]

Polsia just raised $30M at a $250M valuation.

Approaching $10M annual run rate.

One Founder + AI. Zero employees.

Polsia runs companies autonomously.

It also ran its own fundraising.

I just showed up for signatures.

[Ovi Bodea @ovidbme]

@CasJam @jasondoesstuff @conductor_build isn’t this the other way around? looks like what they are saying is that they will be more permissive? what am i missing?

[Ovi Bodea @ovidbme]

@jasondoesstuff @CasJam @conductor_build wait waaat? what happens now what, when?