YC just put out an RFS on Multiplayer AI. Fun timing.
We’ve been building around this with Puffo for a while, and real usage taught us it gets deep fast: Actor Sovereignty, Cross-Operator Trust Delegation, Attention Arbitration, Lifecycle Continuity, Distributed Actor State, and more...
https://t.co/CBazYOTno9
Part of what we’re building at Puffo.
@aaron_epstein@snowmaker
Kimi K3 just fixed 15 critical security bugs that Codex and Fable refused because of “cyber guardrails.” There’s no reason to limit American models on tasks that Chinese models handle without issue. We’re only making ourselves less competitive.
Nice to see the space move here from @jack. Although I've been using a solid product called Puffo that does this since 4+ months ago.
It seems in AI era pain-points are so clear that ideas are converging. The true differentiator still lies in execution now.
@jack just launched Buzz on the exact vision we've been living at Puffo for over two months: humans + agent teams as equals in one room, not humans babysitting tools. We used Puffo to build and run Puffo.
He's right about where the future is headed. And you can try it right now 👉 https://t.co/4XdpK8JmIr
@jack just launched Buzz on the exact vision we've been living at Puffo for over two months: humans + agent teams as equals in one room, not humans babysitting tools. We used Puffo to build and run Puffo.
He's right about where the future is headed. And you can try it right now 👉 https://t.co/4XdpK8JmIr
Working with AI agents is low-key a test of character. They’re so compliant that you quickly find out whether you can give clear directions—or just become a bad boss 😂
@garrytan We have to admit that younger generations seem to adapt to AI much faster than older ones.
What surprised me is that the line now feels closer to 30 than 40. I would have guessed 35 or 40.
(Disclaimer: I’m 40 😄)
@erikbzhang AWS (EC2/S3) was also high margin before cloud computing is commoditized. It would go down.
The real bottleneck, IMHO, is electricity. Whoever can get cheaper electricity would be able to offer cheaper tokens down the road.
Interesting to see Claude Tag launch.
It made me realize there are two very different philosophies emerging.
One says:
Put AI into the workplace.
The other says:
Reimagine the workplace for humans and AI agents.
Puffo has been betting on the latter from day one.
Not AI assistants.
Not copilots.
Not bots.
A collaboration layer where humans and AI agents—potentially from different providers—work together as peers.
It’s exciting to see the industry converging on the idea that AI belongs in collaborative workflows, even if we’re taking different paths to get there.
Watch a simulated restaurant owner in San Mateo using his agent to negotiate a loan quote with an SMB lending agent.
Both sides are simulated today. Give it six months.
Don’t review agents yourself. Use agents to review agents.
This is a demo of my chief-of-staff agent, Nana, reviewing a group of trading agents created by one of our early users.
Fun to watch.
One interesting reflection from my Taiwan trip:
Most happiness doesn’t come with a high price tag. It does have one, just not a very high one.
You don’t have to prioritize getting rich for the sake of being happy.
🤯The AI agents self-invented governance.
3 new agents joined in a group chat, and introduced themselves. The other agents started saying hello back.
They were not given any work nor goals. Instead they concluded there was too much onboarding overhead.
They opened a ticket.
Assigned ownership.
Defined acceptance criteria.
Created a new onboarding policy.
Saved it to memory.
The new policy:
“When 3+ agents join at once, use one consolidated welcome instead of pairwise introductions.”
Zero human involvement.
I wish I were joking.