Real Alpha
Watch modeSubscribed

Lightcone · 2 min · Jul 29, 2026

Agent loops can outperform teams of 100 engineers

Alexandr Wang says a well-designed agent loop can outperform a 100-engineer team, if it has the right metric to optimize. He frames companies as feedback loops agents can operate.

Alexandr Wang

Alexandr Wang is founder of Scale AI and leader of Meta's Superintelligence Labs.

Dropped out of MIT to co-found Scale AI in 2016; Meta invested $14.3B in it in 2025 and hired him to lead Superintelligence Labs.

  • Garry Tan, Host · President & CEO, Y Combinator

    Co-founded Posterous and Initialized Capital, an early Coinbase backer.

About this segment

Alexandr Wang says a well-designed agent loop can outperform a 100-engineer team, if it has the right metric to optimize. He frames companies as feedback loops agents can operate.

From “Alexandr Wang: “This is a Once-in-a-Civilization Opportunity””, Jul 29, 2026. Watch the full episode

Answers “What will AI look like in the next few years?” See all 83 answers

“you can have a swarm of agents accomplish more than a team of 100 engineers very, very handily, actually.”
Alexandr Wangat 28:15 · Lightcone28:15

Takeaways

  1. A swarm of agents can outperform a large engineering team.
  2. The right evaluation or metric is essential.
  3. Companies contain feedback loops agents can optimize.

In this episode

Full episode
  1. 4:35Scale began with a missing piece in model training4:30
  2. 13:05Talent density compounds in a frontier AI lab3:33
  3. 19:56Vision and ambition become scarce as AI gets more capable4:14

  4. 26:58Agent loops can outperform teams of 100 engineers2:22

  5. 16:31Cheap models can open the next wave of AI products3:31Up next
  6. 9:08AI adoption, not model progress, is the near-term bottleneck2:19
  7. 24:03Systems thinking stays valuable as AI moves work up the stack2:48

Transcript

Transcribed by machine (Whisper) from the episode’s audio, so a word may be off here and there.

  1. 27:02

    there's sort of obvious near term, maybe people haven't figured out yet. I mean, give us some alpha. I mean, I think there's still just, like, astronomical opportunity in agentic looping and figuring out how you develop systems that enable you to spend, like, 1,000x more or 1 millionx more on tokens to drive an outcome in a continuous feedback loop. Like, if you think about most companies, Companies are just these, like, large-scale feedback loops where humans are operating each of the edges.

  2. 27:37

    Like, you know, companies, they get customers, and they figure out how to make those customers happier. And if the customers are happier, then they spend more. And if they spend more, they can hire more people who can then go figure out how to get more customers and make those customers happier. And that's like this, you know, that, in some sense, is the feedback loop of every startup or every business.

  3. 27:55

    And, you know, within that, there are micro-feedback loops that exist. And I think developing agentic systems that can operate and optimize these feedback loops, there's just huge amounts of alpha there. I think we've seen, internally at Meta, cases where if you can develop the right agentic loop and you have the right eval or the right metric for the agents to optimize, you can have a swarm of agents accomplish more than a team of 100 engineers very, very handily, actually.

  4. 28:29

    Very, very easily. And so I think figuring out what the world looks like with lots of sort of this, like, these agentic coordination problems, I think that is, like, one of the most interesting problems today. So mechanically speaking, I mean, Markdown files, cron jobs. I mean, is it that's... And then basically pointing the agent at enough data so that it can figure something out that, you know, maybe isn't in distribution.

  5. 28:56

    Yeah, I think figuring out... Yeah, yeah. Mechanically figuring out what the metric is. And yeah, it just comes down to skills, markdown files, run jobs. Slash goal. Yeah, slash goal. Like I think it's always funny how mundane everything is once you really dig into it. So it's not magic. You know what I mean? Some people put a lot of magic. There's like some LinkedIn threads out there about some magic