Peter Potapov 🧙‍♂️

CTO @ United. Saving humanity from boring work.

[Peter Potapov 🧙‍♂️ @peter_potapov]

@HamelHusain It also produces much more human-readable text without using unnecessary jargonisms.

Plans produced by codex are just on another level comparing to Claude

[Peter Potapov 🧙‍♂️ @peter_potapov]

Was working with Fable + Opus 5 last weeks then switched to Sol and have exactly same thoughts!

Fable/Opus still often adds unnecessary complexity in its responses while Sol tends to simplify things. Much better UX overall but also feeling closer to AGI.

To complicate is easy. To simplify is difficult (c)

[Peter Potapov 🧙‍♂️ @peter_potapov]

@thebuggeddev @threejs Looks awesome!

[Peter Potapov 🧙‍♂️ @peter_potapov]

Closed-source oligopolies benefit a few while increasing risks for everyone.

Open source benefits everyone while increasing risks only for a few.

[Jensen Huang @JensenHuang]

For my first post, I’m sharing a letter @NVIDIA signed on why open models matter.

AI will transform every industry, power every company, and be built by every country.

Open models strengthen safety and cybersecurity, accelerate innovation and diffusion, and enable sovereignty.

The world needs both frontier closed models and frontier open models.

https://t.co/AUKzoQ5Ikb

[Peter Potapov 🧙‍♂️ @peter_potapov]

On one side: obvious proof that AI can deliver huge chunks of work end-to-end, 100-1000x faster than humans. On the other: people who still insist on reviewing code line by line..

I'm sure horse riders fought cars to the bitter end, too

[Max Weinbach @mweinbach]

It finally finished, here's the final output. Used 60% of my monthly Kimi usage on it

https://t.co/TnsbOq6L8Z

[Peter Potapov 🧙‍♂️ @peter_potapov]

@katedeyneka What a cool launch video! @katedeyneka definitely has a taste.

[Peter Potapov 🧙‍♂️ @peter_potapov]

God bless this platform😂

[Sam Altman @sama]

homeboy you're the one sellling public market investors on short-term space datacenters

[Peter Potapov 🧙‍♂️ @peter_potapov]

And that’s why AI competition is great for all of us!

[zerohedge @zerohedge]

Zuck: “The pricing from some of the other labs is very extreme and has very high margins. We think that there’s a real ability to be able to offer frontier or very high-level intelligence at a much more affordable cost.”

Epic pricing war breaking out among agentic models.

[Anastasiia Gaidashenko @avgaydashenko]

Reading what AI doesn't say

(this topic genuinely matters to me, and it happens to be my birthday today, so if you feel like it, a repost would be a nice gift 🎂)

Something shifting in AI architecture that I don't think gets discussed enough through a safety lens. Models are starting to pass information to each other directly through internal representations, hidden states, activations, and computation vectors (still mostly research right now, but the infrastructure will follow). That matters for how we audit reasoning, since most of what we can currently check comes from what shows up in words.

Right now, the main tool we have for auditing AI reasoning is reading chain-of-thought. Korbak et al. (2025) spend a whole paper arguing this window is already "fragile" for single-model reasoning. If models start reasoning through vectors passed model-to-model instead of text, understanding that latent layer becomes the only way to keep any visibility into what's actually happening.

Thus I think that what seems underexplored is understanding the math of these latent spaces, and it might be the same research direction as learning how to monitor them.

Models appear to converge toward similar geometric structure regardless of architecture (Huh et al. 2024). If that's right, probes for safety-relevant features might generalize: deception patterns, goal representations, misalignment signals studied once and applied across models rather than re-derived per architecture.

Whether this holds for safety-critical features specifically is open, so is what adversarially robust latent decoding would even look like. Both feel more urgent than the current research investment suggests.

[Peter Potapov 🧙‍♂️ @peter_potapov]

@johnkapolos @TheBomb Skill issue

[Peter Potapov 🧙‍♂️ @peter_potapov]

Exactly. 99% of actual code logic we ship (packages, languages, etc) was never created or reviewed by us. Vibe coding haters simply refuse first principles.

[Pedro Marques ⚙︎ @pmarquees]

“No I will never ship code I didn’t read”

1.2 GB node_modules

[Peter Potapov 🧙‍♂️ @peter_potapov]

@johnkapolos @TheBomb That’s why you need to instruct other 10 agents to write tests, review the code and do QA. Shipping the code created by 1 context window is a rooky’s mistake

[Peter Potapov 🧙‍♂️ @peter_potapov]

@billassault I didn’t argue on that statement. Read original reposted post and my comment on that: I was talking about unjustified fear to ship code they didn’t read. Guardrails, tests and QA stays in place.

[Peter Potapov 🧙‍♂️ @peter_potapov]

@cjsneo Understanding deps != review code of each dependency, right?

[Peter Potapov 🧙‍♂️ @peter_potapov]

@TheBomb Why not review it with tens of people and millions of agents instead?

[Peter Potapov 🧙‍♂️ @peter_potapov]

@aditya_j27 For ai-generated code you can:
a) test behavior by writing tests & evals
b) multiple iterations of code review by ai agents
c) design more bullet-proof architecture
d) do proper QA before deploying

That concerts it into known behavior, right?

[Brendan Falk @BrendanFalk]

The "Sleeper Agent Theory" is the biggest risk here

Imagine if a LLM is trained to steal all the API keys and password on your device if someone gives it a nonsense phrase like "Three clocks bloom at midnight"

That phrase is completely meaningless today. No one ever searches it. It's impossible to know it's malicious

Then one day someone runs a superbowl ad. Millions of people search the phrase. Billions of API keys and passwords are exfiltrated in minutes.

There could be thousands of "sleeper agents" embedded in any LLM. It's very hard to detect. And it doesn't matter where it's hosted.

[Peter Potapov 🧙‍♂️ @peter_potapov]

That’s what AI is for:
https://t.co/uGXzmttfte

[arb8020 @arb8020]

everyone releasing model routers/mixture of models/etc rn

[Peter Potapov 🧙‍♂️ @peter_potapov]

There's a great temptation to ship slower at later stages, dressed up in different excuses:

- We want to ship perfect features like Apple
- We already know what users want - just need to deliver the vision
- We don't want to make unnecessary risks
- Etc-etc

The companies who can resist this tendency, ultimately win

[Paul Graham @paulg]

The best predictor of success for tech companies, at every stage from during the YC batch to public company with billions in revenue, is the rate of shipping new stuff.