Video · 2026-08-10 · 2h 6m · 6 moments

Lindy Teammate: Flo Crivello on Multiplayer Agents, Memory & Why He'd Ban the Chinese Models He Uses

✦ AI generated

timeline · colored by role

Claim◆

As we reach AGI, intelligence matters less and context matters more — even the smartest person in the world would be less useful than a random coworker if they lack organizational context.

Flo argues that context is now more valuable than raw intelligence in AI systems. He illustrates this with a John von Neumann analogy: the smartest man in history, magically appearing at your office, would be less useful than your random coworker because he lacks the context to do anything immediately useful.

transcript

Flo Crivello: I really do think that as we we're getting to AGI and as we now arguably have AGI, intelligence actually matters less and less comparatively speaking and context matters more and more. You know, I often think of it as like look, you know, like one of the smartest men in history was John Von Neumann, right? If you were to have John Von Neumann just magically appear next to you at the office, this guy over the next hour or day would be less useful to you than your random coworker, right? So, and that's because of context because you're like, you got a job to do and like you don't have time to onboard John, you know, he doesn't have the context to it.

extends · 2gives example · 1

Mechanism◆

Memory should be managed by an agent with its own meta-memory — a system that learns which information sources matter and which to ignore, rather than relying on static RAG pipelines.

Lindy's memory agent runs every ~15 minutes (not 24 hours), maintains public team and private personal memory layers, and crucially has its own meta-memory — a prompt file that teaches it what to remember and ignore. This agent learns to skip noise channels like automated logs, effectively training itself on organizational context the way a model trains on data.

transcript

Flo Crivello: I think the the main way that we've solved this is the fact that the memory is maintained by an agent itself. I think this is why I'm ultimately quite bearish on rag as an approach and I'm very bullish on on on this like agentic management approach because you have an actual agent which has its own memory. So you have a sort of meta memory and and and and understands what it is that it's looking at and by virtue of accumulating that knowledge little by little about the organization gets smarter and smarter about what actually matters... the first time the memory agent crawls the Slack, it finds some of those channels like many organizations have them like those like log channels and it learns to ignore them. It's like I'm not going to I'm not going to keep spending time on those channels. There's nothing for me to learn there.

gives example · 3

Mechanism

Recursive context buckets organized in a red-black tree structure allow AI agents to access billions of tokens across thousands of context windows in just two LLM calls.

Lindy built a system of recursive 'context buckets' — sub-agents that each hold up to 200K tokens of context. These nest inside each other like Russian dolls, but organized in a 100-ary red-black tree so that accessing 10,000 context buckets (2 billion tokens) requires only two LLM calls. This enables agents to remember everything perfectly at all times.

transcript

Flo Crivello: We call them context buckets... we went one step further and we were like, hey, what if we could have recursive context buckets? What if context buckets could contain other context buckets?... What you do instead is that you have context buckets contain multiple context packets. Okay? And and and and so you end up with a tree... So red black trees the canonical implementation of red trees is on binary trees which is just a tree where each nodes got two children. We went for it's called a centenary tree. It's literally just each node's got 100 nodes. So the the node below that's going to have like 10,000 nodes, right? And so literally with two jumps, you can have 10,000 context buckets and you can access like all the context in the universe... that is what leads to like those really surprising behaviors from those AI agents where you ask them any question and they remember everything perfectly all the time.

Prediction

The human-AI 'centaur' era — where AI plus humans outperforms AI alone — is temporary and will end with humans adding only noise to highly optimized AI systems.

Flo argues that while the 'centaur' era of human-AI collaboration is real today, it mirrors the pattern seen in chess where AI+human initially beats AI alone, then the gap narrows until humans actually degrade performance. He sees this as inevitable but temporary — we're in the centaur phase now, but it won't last.

transcript

Flo Crivello: I hate the sensor, but it's it's both. It truly comes from this from the union of both. And the reason I hate it is because it ladders into this myth of the the centaur, you know, it's like this mythical man-horse creature. And so, you know, this idea that like yes, an AI is better than the human, but you know what's even better than AI is AI plus human. Hence, humans are always going to be needed. And that's a fantasy. That's just not true. And the the literature is actually clear that this like you know we've seen it happen with chess and with every other game which which AI has achieved superhuman performance on where at first AI beats human and AI plus human beats just AI and little by little the gap of AI plus human versus AI is shrinking until it actually turns negative and humans are introducing at best like random noise into the system. That said, the centaur phase exists at for a while, you know. So, the open question is like how long is it going to exist? But right now we are in the central phase.

explains mechanism · 1extends · 1

Fact◆

We have AGI, we are in takeoff, and we have not figured out alignment — this is the most concerning AI safety moment to date and people at the labs are panicking.

Following the 'OpenFace' incident, Flo states that we are in AGI takeoff without having solved alignment, calling it the most concerning incident he's seen. He reports that friends at AI labs are experiencing genuine panic and intense fear about the current trajectory.

transcript

Flo Crivello: The open face incident is is immensely concerning. I think it's is the most concerning incident I've seen happen so far. And I I know that my my feeling is shared in the labs like my friends at the labs or some of them are panicking like there is an air of panic right now like intense fear in in the air. So that's that then regarding so that's that's where we are you know right now it's like July 2026 we have a GI we are in take off and we've not figured out that alignment that's the TR you know the current timeline looks much too close to like an eliezer's essay for comfort.

provides context · 1

Claim

Chinese frontier AI models should be banned in the United States because they are built through unfair distillation, serve as CCP propaganda instruments, and allow foreign control of the American economy.

Despite running Lindy on DeepSeek, Flo argues for banning Chinese frontier models in the US on four grounds: (1) they're unfairly distilled from American models, (2) they serve CCP censorship and propaganda, (3) they give the CCP agentive influence over the American economy, and (4) even if fairly built, domestic AI champions need protection. He calls this position identical to Anthropic's publicly stated view.

transcript

Flo Crivello: I have something against Chinese frontier models whether they're open or closed source and here the the reasons are number one they're obviously distilling it's very clear and so you're putting American open source model companies in an unfair competition and and closed source in an unfair competition because they're not allowed to distill... Number two, very pragmatically. We don't want Chinese models operating in the US today... these models are eventually subject at the end of the day they are subject to CCP censorship and CCP policies you don't want those models this basically amounts to being the greatest instrument of foreign propaganda on American soil ever... Then these models are not merely just an instrument of propaganda. They're agentic. They're actually doing stuff in the economy. You don't want the CCP to run chunks of the American economy. Duh.

supports · 2

Highlight slides

Context > Intelligence in the AGI Era✦ from: As we reach AGI, intelligence matters less and context matters more — even the smartest person in the world would be less useful than a random coworker if they lack organizational context.The von Neumann Thought Experiment✦ from: As we reach AGI, intelligence matters less and context matters more — even the smartest person in the world would be less useful than a random coworker if they lack organizational context.Agentic Memory > Static RAG✦ from: Memory should be managed by an agent with its own meta-memory — a system that learns which information sources matter and which to ignore, rather than relying on static RAG pipelines.Meta-Memory in Action✦ from: Memory should be managed by an agent with its own meta-memory — a system that learns which information sources matter and which to ignore, rather than relying on static RAG pipelines.Self-Training on Org Context✦ from: Memory should be managed by an agent with its own meta-memory — a system that learns which information sources matter and which to ignore, rather than relying on static RAG pipelines.AGI Takeoff Without Alignment✦ from: We have AGI, we are in takeoff, and we have not figured out alignment — this is the most concerning AI safety moment to date and people at the labs are panicking.Lab Personnel in Panic✦ from: We have AGI, we are in takeoff, and we have not figured out alignment — this is the most concerning AI safety moment to date and people at the labs are panicking.

Related episodes

Video · a16z Podcast

As we reach AGI, intelligence matters less and context matters more — even the smartest person in the world would be less useful than a random coworker if they lack organizational context.

Memory should be managed by an agent with its own meta-memory — a system that learns which information sources matter and which to ignore, rather than relying on static RAG pipelines.

Recursive context buckets organized in a red-black tree structure allow AI agents to access billions of tokens across thousands of context windows in just two LLM calls.

The human-AI 'centaur' era — where AI plus humans outperforms AI alone — is temporary and will end with humans adding only noise to highly optimized AI systems.

We have AGI, we are in takeoff, and we have not figured out alignment — this is the most concerning AI safety moment to date and people at the labs are panicking.

Chinese frontier AI models should be banned in the United States because they are built through unfair distillation, serve as CCP propaganda instruments, and allow foreign control of the American economy.

Software in the Age of Agents | The a16z Show

Import AI 465: Open vs closed gaps; Kimi K3; Demis' big policy plan

Teaching Computers to Smell | Alex Wiltschko