26 chunks · 11 episodes
Human judgment will always need to be marbled into agent swarms. The human judgment grounds the swarm, making sure the output has meaning. The leverage factor (agent to human ratio) might go up and change, but th...
Agent Swarms feel like Facebook culture circa 2008. Hire a swarm of driven but inexperienced 20-somethings, give them a very general target to sight off, and let...
Agent swarms are empowering... and anxiety producing!
...at there's a natural (low!) ceiling for things above the API. When you move the agent swarm below the API, you can now have many more items without hitting that scaling factor. The limit now becomes how many tokens you want to invest.
...lying, having to touch the ground is excruciating. When you're interacting with agent swarms, you're flying. A very small amount of interaction creates huge amounts of value. When you have to talk to a human, you're grounded. The leverage is...
...nts expand the coordination cost super-linearly. That's one of the reasons with agent swarms, dangerously-skip-permissions is so freeing. It's also freeing in human organizations, with "Ask forgiveness, not permission." You trade off executi...
...sume Compromise. Agents of Chaos: a paper on directly observed failure modes of agent swarms in the lab.
With agent swarms it's possible to be out over your skis and not realize it. Normally when interacting with other humans, someone will point out that you might be wro...
Is the LLM model above or below the API in your system? Is the agent swarm above or below the API? Does a user have to think about it?
A friend who conducts multiple agent swarms recently had his Claude Max account turned off by Anthropic. He wasn't doing anything against the Terms of Service… and wasn't even using Claude as ...
...tures that work best for human organizations don't necessarily map to the ideal agent swarm organizations.[ao] Human structures have to be resilient to humans' impatience, difficulty to focus, and emotions. It took us centuries of trial-and-...
...ation. What about human-to-human cooperation, now facilitated and lubricated by agent swarms? Agent swarms as roller bearings to reduce friction between human systems.
...ut an upcoming paper that digs into optimal organizations by experimenting with agent swarms. If you have an even number of agents, the swarm will sometimes get locked in a tie and plateau. Organizations with an odd number of agents keep get...
...e up to replace them. Could you implement an idea lab structure built out of an agent swarm? It seems intuitively like you perhaps could not… the agents can't die, so they have no existential draw to the ideas they randomly espouse.
It's anxiety producing to lead a synchronous agent swarm. The chat modality makes you feel social anxiety for making them wait. A swarm of agents, chirping for attention. An async interaction model is more ...
If you have a clear performance metric to optimize, agent swarms can do a great job. This is what Shopify's Tobi found. Interestingly, in his case, there weren't any magic bullets. It was just an accumulation of t...
...Synchronizing with external stakeholders about the thing you're working on with agent swarms is excruciating. You don't need to sync with your agent swarm, they just fly. The other human you're coordinating with can ask questions that are al...
... unleashes the power of their insight to an extraordinary degree. In a world of agent swarms, the limiting factor is your judgment and intelligence.
20 minutes of well-orchestrated agent swarms is equivalent to a day of an amazing team
...ut leisure in "Economic Possibilities for our Grandchildren". Everyone who does agent swarms leans towards empowerment to the point of mania. This is due to the extreme opportunity cost. The opportunity cost of a minute is now insane. Which ...