Intriguing article. There's a little editorial error; the sentence "an AI legislature negotiating over scarce resources produced a 10,000-word constitution and almost no policy" appears twice.
That's a compelling topic to explore. One thought came up while I was reading it. Large-scale collaboration can only happen through stories. Money, religion, governments, corporations, and my other ideas are stories we created and keep telling each other.
What story should we tell to make this vision work?
Timour — "the intelligence is in the group; it just does not reliably become action" is the sentence that reframes agents from productivity tools to infrastructure. That's a fundamentally different design problem and most builders aren't thinking about it yet.
I build custom AI agents and the coordination use case is the one that excites me most while also being the hardest to sell. A solo business owner understands "answer my DMs at 10pm." A community understanding "ambient intention matching across 500 rotating participants" is a much longer conversation. But you're right that the leverage is orders of magnitude higher. An agent that helps one person save two hours is nice. An agent that helps thirty people find the three who share their project goal and actually follow through — that's a different category of tool entirely.
The historical lineage is the smart framing. Double-entry bookkeeping didn't just help merchants track money — it enabled the joint-stock company. Standardized containers didn't just move boxes — they restructured global trade. The question is what new organizational form agents enable that we can't see yet because we're still thinking of them as personal assistants.
The Edge Esmeralda experiment is the right move — theory without testing is a pitch deck. Curious what you learn about follow-through facilitation specifically. That's where every coordination tool I've seen breaks down — the gap between "we agreed" and "someone actually did it."
Agents don’t really remove coordination work. They just stop letting us pretend it wasn’t there. The funny part is that many teams will plug agents into broken workflows and then be surprised when they get faster broken workflows.
The winners won’t be the firms with the most agents. They’ll be the ones clear enough about the work that an agent can actually help.
Timour, you make a persuasive case for the role of agents in fostering 'ambient' or shared social and democratic interests, which is something to look forward to. But there's a whole other dimension to this same technology that might temper our enthusiasm — the likelihood that commercial actors can and will use it in less productive and desirable ways.
“we have no idea what effect this will have on society”
I don’t think that’s quite right. There’s ample science fiction examples to pull from. Snowe crash, the culture series Vernon vinge etc. As humans we also have the capacity to imagine and think through different scenarios and also how they might align with goals, values and beliefs. History also provides parallels and examples to learn from. Experiments like what you are conducting are important and it’s a real frontier albeit not completely alien terrain.
Thanks for this well researched and thought out article!! I learned a lot!
Thank you, Michel. This means a lot coming from you!
coming from a sincere and admirative heart <g>
Intriguing article. There's a little editorial error; the sentence "an AI legislature negotiating over scarce resources produced a 10,000-word constitution and almost no policy" appears twice.
banger
I have a similar obsession with coordination problems! They are a problem as old as time but I think a new category of coordination problem will emerge as a consequence of how AI shifts economic organization: https://altitudewrites.substack.com/p/everything-starts-and-ends-with-human?r=2pnvl&utm_campaign=post-expanded-share&utm_medium=post%20viewer
100% agree, and I'm excited to read this :)
That's a compelling topic to explore. One thought came up while I was reading it. Large-scale collaboration can only happen through stories. Money, religion, governments, corporations, and my other ideas are stories we created and keep telling each other.
What story should we tell to make this vision work?
Timour — "the intelligence is in the group; it just does not reliably become action" is the sentence that reframes agents from productivity tools to infrastructure. That's a fundamentally different design problem and most builders aren't thinking about it yet.
I build custom AI agents and the coordination use case is the one that excites me most while also being the hardest to sell. A solo business owner understands "answer my DMs at 10pm." A community understanding "ambient intention matching across 500 rotating participants" is a much longer conversation. But you're right that the leverage is orders of magnitude higher. An agent that helps one person save two hours is nice. An agent that helps thirty people find the three who share their project goal and actually follow through — that's a different category of tool entirely.
The historical lineage is the smart framing. Double-entry bookkeeping didn't just help merchants track money — it enabled the joint-stock company. Standardized containers didn't just move boxes — they restructured global trade. The question is what new organizational form agents enable that we can't see yet because we're still thinking of them as personal assistants.
The Edge Esmeralda experiment is the right move — theory without testing is a pitch deck. Curious what you learn about follow-through facilitation specifically. That's where every coordination tool I've seen breaks down — the gap between "we agreed" and "someone actually did it."
Great piece! I can't wait for the results from the Edge Esmeralda village experiment!
Agents don’t really remove coordination work. They just stop letting us pretend it wasn’t there. The funny part is that many teams will plug agents into broken workflows and then be surprised when they get faster broken workflows.
The winners won’t be the firms with the most agents. They’ll be the ones clear enough about the work that an agent can actually help.
Timour, you make a persuasive case for the role of agents in fostering 'ambient' or shared social and democratic interests, which is something to look forward to. But there's a whole other dimension to this same technology that might temper our enthusiasm — the likelihood that commercial actors can and will use it in less productive and desirable ways.
This is such a pleasant write. I want to see films made about the next frontier of human coordination mentioned in this blog!!
Agentic layer as a dark matter to our human matter universe :p. Great one def
“we have no idea what effect this will have on society”
I don’t think that’s quite right. There’s ample science fiction examples to pull from. Snowe crash, the culture series Vernon vinge etc. As humans we also have the capacity to imagine and think through different scenarios and also how they might align with goals, values and beliefs. History also provides parallels and examples to learn from. Experiments like what you are conducting are important and it’s a real frontier albeit not completely alien terrain.