The alpha, twice a week
That was 2 min instead of 1 h 5 m.
Get the best new moments in your inbox every Wednesday and Saturday. It's free.
Check your inbox
We sent a link to your address. Click it to finish signing up. Not there in a minute? Look in spam.
or
The Knowledge Project · 2 min instead of 1 h 5 m · Sep 15, 2026
Machines can't take responsibility, so use an AI council
Tobi Lütke
Founder & CEO, Shopify
Shane ParrishHost · Founder, Farnam Street
Former intelligence agency employee; wrote the bestseller Clear Thinking.
About this segment
Tobi says AI builds the environment for judgment but cannot own decisions. For big calls he runs a council of sub-agents across several models, randomizes the synthesizer, and spends about $15 to $20.
From “Tobi Lütke: The Skills AI Makes More Valuable”, Sep 15, 2026. Watch the full episode
“Here's the thing that LLMs and machines cannot do. Machines can't take responsibility.”
Takeaways
- Humans must own the decision; AI informs it
- A council of expert sub-agents runs across several different models
- The synthesizer is randomized, and a run costs about $15 to $20
In this episode
Full episode- 36:35Intuition matters most when the best path has no feedback1:44
- 40:39The hard part is choosing among many good options1:55
14:02Machines can't take responsibility, so use an AI council2:03
Transcript
Transcribed by machine (Whisper) from the episode’s audio, so a word may be off here and there.
- 14:05
use it for. Like I use it for like creating the right environment for judgment. Here's the thing that LLMs and machines cannot do. Machines can't take responsibility. And I think this is actually probably the most overlooked thing in the entire stack. Humans take responsibility. Machines can help us take more responsibility because they can inform us better. Like this is what a dashboard does. You know, the word of Wall Street traders knows this very well.
- 14:36
You get yourself a perfectly set up Bloomberg terminal to make decisions, but like you have to make a call, right you can't make it make the call creating human in the loop decision surfaces is a way to i think describe the ideal environment if i need a really really really important decision made and i really need an exceptionally good like give me the most neutral ground truth you know then what happens is a small little council is created of five six different experts like one is data role. One is like do paper research. One is like the business perspective. One is maybe engineering perspective on a thing. We're running this sub-agent, like my thing runs a sub-agent then, you know, against like, you know, Grog, ChatGPT, Opus, and maybe Kimi now. That changes all the time. It runs each of them against each of his models. Then there's a synthesis step
- 15:36
where it's randomized who is synthesizing the thing. Synthesis is all pooled. All of that is being read usually by the best model that exists right now. This would be like Fable. That's the conclusion that comes back to me. And you spend 15, 20 bucks on tokens. But you get something. in like half an hour which is like you could have also done but you could spend a month on it has