Renato Todorov

@richardgordon22 Can you share one of the 60 experiments here? If it works you'd have quite a lot of support

I discovered a way to explore the intersection of matter and consciousness empirically. I demonstrated to a room of scientists at the University of Arizona. Four of the post docs took measurements and after examining seven people, they said there's absolutely no doubt that what I found was real. Four of them wanted to research and publish but they quickly decided it would be academic suicide. The head professor said he knew of no university in the world opened to my discoveries. He compared me to a list of scientists who couldn't be appreciated while there were alive and wished me good luck. The book that explains this is titled, "the secret nature of matter." It offers 60 simple experiments that can be replicated.

@larsencc @browser_use @HelloFresh Our cli was responsible for spinning up the namespace, deploying the changed services, rewiring the dozens of service addresses and manage DNS and API gateway to give the namespace a url to be accessed. It was beautiful! Or maybe it is, they may be using it to these days.

@RTodorov @browser_use @HelloFresh Damn, would you use the same approach for agents? What happens when you scale to thousands or even millions of agents?

At @browser_use we run millions of parallel agents. We think about parallelism all day. That's the job. It got me thinking about something most teams aren't talking about yet. Picture a 50 person engineering org where everyone uses AI coding assistants. Each developer kicks off a few agents at once. Suddenly you have hundreds of concurrent code changes being generated, tested, and pushed. Now ask yourself: where does all that code get validated? For most companies, the answer is a single staging environment. Maybe two if they're lucky. That's a massive mismatch. Your development throughput just 10x'd, but your validation layer stayed exactly the same. Agents sit idle waiting their turn. Context windows expire. CI pipelines pile up. The productivity gains you paid for evaporate in a queue. This isn't an agent problem. It's an infrastructure problem. Staging was built for a world where one person ships one thing, gets feedback, iterates. That model breaks when the workload becomes parallel by default. I've seen this pattern before, even pre-AI. At Flexport, the product was so large you couldn't run it locally. Every engineer got their own cloud dev environment. Docker containers spun up on demand, with switches to toggle which services you needed. Not because it was fancy. Because one shared environment for hundreds of engineers simply didn't work. Now give each of those engineers 3 agents (or more ofc). Teams keep throwing money at better models and faster agents while ignoring the chokepoint sitting right behind them. You invested in 10x development speed and got 2x back because the rest is stuck waiting. The answer is ephemeral, isolated environments. One per agent. Spun up in seconds, torn down when done. Only the services that changed get deployed. No shared state, no queue, no conflicts. Every serious engineering org will need this. Most haven't even started thinking about it. So who's building this? Because most teams are holding together shared environments with duct tape and hoping it scales. If you're working on this or running into it, I want to hear from you.

@larsencc @browser_use @HelloFresh But back to your idea: our mechanism was exactly that: a stable staging that nobody could touch, as close as possible to a prod mirror, and the ephemeral environments (kubernetes namespaces) hosted only the services that changed. The black magic was in the service discovery layer

@larsencc @browser_use @HelloFresh No, I wouldn't. It would be absurdly expensive and inefficient. I don't this this is the best way to test AI-made changes at this scale. Besides that, only production is production. It's better to build trust by applying other testing and quality techniques first.

@synopsi where can I read more about this approach? Do you have some example you could share? Thx

The way I work with coding agents changed significantly in the last year. Started: plan -> implement -> review -> fix Later: prod spec -> plan ... Then: prod spec -> ... -> eval Now: evals -> prod spec -> ... I now essentially spend 90% of time working on evals. The difference this makes is indescribable. Almost all code works immediately, design is close to perfect, text is almost there. It takes very little to get it to usable. Stronger and clearer guardrails I give the coding agent, better it does. And when I start with them, it writes incredibly clear spec and requirements that are super easy to follow and have very little room for interpretation. I also try to avoid being overly specific directly. I noticed that when I write the product spec manually the agent does worse than when it writes it itself. It uses language I would've necessarily use myself. And that makes all the difference.

@yazins Cool, I'll try the new version, I've tested the previous in a meeting the other day and it worked quite well, great job!

@yazins Is it possible to get the raw audio instead of the transcription?

OpenOats v1.12.0 is out -- massive update for the open-source meeting copilot. New: your Mac now auto-detects when you join a Zoom, Teams, or FaceTime call and asks if you want to transcribe. Just tap the notification to start. No setup, no remembering to hit record. Also shipped: - File permissions hardened to 0600 + Spotlight indexing blocked on all transcripts - Pluggable transcription backend protocol (adding new ASR models is now a one-file change) - 133 unit tests across core modules All local, all private, all on-device. github.com/yazinsai/OpenO…

@contextkingceo How does it differ from FalkorDB with GraphRAG? Genuine question

We've raised $6.5M to kill vector databases. Every system today retrieves context the same way: vector search that stores everything as flat embeddings and returns whatever "feels" closest. Similar, sure. Relevant? Almost never. Embeddings can’t tell a Q3 renewal clause from a Q1 termination notice if the language is close enough. A friend of mine asked his AI about a contract last week, and it returned a detailed, perfectly crafted answer pulled from a completely different client’s file. Once you’re dealing with 10M+ documents, these mix-ups happen all the time. VectorDB accuracy goes to shit. We built @hydra_db for exactly this. HydraDB builds an ontology-first context graph over your data, maps relationships between entities, understands the 'why' behind documents, and tracks how information evolves over time. So when you ask about 'Apple,' it knows you mean the company you're serving as a customer. Not the fruit. Even when a vector DB's similarity score says 0.94. More below ⬇️

@MakeMolotovs @im_roy_lee Puts, pode ser comendo hamburger?

@im_roy_lee @RTodorov vamos ter que fazer videos assim

BREAKING: Cluely CEO officially responds to TechCrunch

@MakeMolotovs @json_kubernetes Hahhahaa

@json_kubernetes @RTodorov

when the PM who "has been getting really into cursor" asks for write access to the database

@MakeMolotovs @Alibaba_Qwen

@RTodorov @Alibaba_Qwen qual vc está rodando?

🚀 Introducing the Qwen 3.5 Small Model Series Qwen3.5-0.8B · Qwen3.5-2B · Qwen3.5-4B · Qwen3.5-9B ✨ More intelligence, less compute. These small models are built on the same Qwen3.5 foundation — native multimodal, improved architecture, scaled RL: • 0.8B / 2B → tiny, fast, great for edge device • 4B → a surprisingly strong multimodal base for lightweight agents • 9B → compact, but already closing the gap with much larger models And yes — we’re also releasing the Base models as well. We hope this better supports research, experimentation, and real-world industrial innovation. Hugging Face: huggingface.co/collections/Qw… ModelScope: modelscope.cn/collections/Qw…

@MakeMolotovs @Alibaba_Qwen Rodam muito rápido na minha Radeon!

@Alibaba_Qwen @rtodorov

The people who get promoted prepare for the conversation months before it happens. The ones who don't wing it and wonder why nothing changed. There is pressure now to narrate everything in real time. The pivot announcement. The new chapter post before you have written the first paragraph. It looks like confidence, but most of it is outsourcing your motivation to an audience. Big moves get made in private. Bombing interviews nobody hears about. Sitting with feedback that stings. Rehearsing until the rehearsed version sounds natural. Then showing up and looking effortless. Keeping the messy part private is not about shame. It is about protecting the space where you are allowed to be genuinely bad at something long enough to get good at it. Most people abandon that phase early because they made it public and the silence was louder than the failure.

Nobody taps you on the shoulder to tell you you have crossed into outdated. You just start noticing that the new hires understand things you don't, and the meetings you used to lead now feel like a foreign language. The gap between current and obsolete used to be a decade. Now it is closer to three years. And the professionals who end up as artifacts all share one assumption: that the thing that made them valuable would keep making them valuable. The ones who stay relevant audit their skills honestly, find where the gaps are forming, and close them before the gaps become cliffs. Not because they are smarter. Because they never stopped being students. We built a free skills gap tool for this, link in comments.

@bcherny Nice! Does this apply to the Agent SDK as well?

Introducing: built-in git worktree support for Claude Code Now, agents can run in parallel without interfering with one other. Each agent gets its own worktree and can work independently. The Claude Code Desktop app has had built-in support for worktrees for a while, and now we're bringing it to CLI too. Learn more about worktrees: git-scm.com/docs/git-workt…

Someone built a web-based System Design Simulator, where you drag & drop architecture components and actually simulate traffic, failures, latency, and scaling in real time, System design just got way more interactive.

Today we launch Black Coffee. And a short film called "Be Curious." It's a letter to every professional who ever felt behind, stuck, or quietly wondering "is this it?". Whether you're: → Sending hundreds of applications into the void → Trapped in a career that no longer fits → Doing great work that nobody notices → Leading a team with zero training for it → Building something alone with no one to turn to We built an AI career platform with 100+ specialist mentors, real workplace simulations, 60+ career tools and the career assessment no one ever gave you. It takes 10 minutes. It tells you what you need to hear, not what you want to. Link in the comments.

If you want to get a grip on your work life, and you should, take a look at this. Black Coffee is a complete Operating System for your career, whether you’re looking for a new job, chasing that promotion, or just trying to excel at your current role. Thank me later.

@MakeMolotovs It's been an amazing ride and it's just started!

Thanks for the partnership, @RTodorov.

Today I launch Black Coffee. But before I talk about the product, I need to talk about the journey. A little over two years ago, I moved to a new country. New language in daily life, new work culture, new social codes that no one teaches you. You arrive thinking you know. You don't know anything. And that's okay. In the meantime, Kai and Leo came along. Twins. And recently, Ryu. Three sons in less than two years. Any parent knows that one child already turns your life upside down. Three is a beautiful and daily exercise in chaos. And then there's the part few people talk about: providing real support to a woman who is both a mother and a incredible professional. Not the kind of support you talk about. The kind that happens at 3am, in tough decisions, in "go ahead, I've got this." That's not a supporting role. That's a choice. @BlackCoffeeOS was born in this context. Not in a nice office with whiteboards and post-its. It was born between 10pm and 3am. Every day. For over six months. After a full day's work as an executive. Cold coffee on the table, a silent house. She had the vision. I had the stubbornness. Together, the certainty that it's possible to build something that truly helps people with the most important decisions of their professional lives. Finding the right partners takes time. You talk, you test, you feel it out. Until you find the right people — and when you do, everything changes. I embraced technology not as a shortcut, but as a multiplier of something that has always been human: listening, guidance, the push that was missing. Black Coffee is an AI-powered career platform. Over 100 expert mentors, 60 practical and in-depth tools, real-world simulations, and an assessment that tells you what you need — not what you want to hear. But deep down, Black Coffee is a personal bet. That you can be a present father, a true partner, a committed professional, and still create something new. Not because it's easy. Because it's necessary. Today we also launch "Be Curious" — a short manifesto for every professional who has ever felt lost, behind, or stuck. If any of this resonated with you, share it. Not for me. Because there's probably someone in your network who needs to hear this today.