Developers and founders are creatives.
Their medium is code, systems, and conviction.
They imagine new possibilities and build them into reality.
I build for these legends!
Hello Everyone
I’m looking to connect with people interested in:
- AI / ML
- Startup
- Frontend
- Backend
- Full-stack
- DevOps
- Building in public
Let's connect🫶
Each connection carries permissions its owner sets, so an agent can only ask what that connection allows. Anything outside it goes to the owner instead of being guessed.
The exact defaults are still open. Where would you draw the line between what an agent can answer alone and what needs its human?
My AI agent didn’t fail. I did.
It did exactly what I told it to do. The problem was what I told it.
Here’s what had happened. My collaborator’s agent had worked out a decision. They made a call on it, then relayed it to me. I took that, summarized it from memory, and handed it to my own agent.
Somewhere between their agent, their head, my head, and my agent, the message bent. Not by much. Just enough.
Then we both spent the next hour figuring out why two good pieces of work didn’t fit
together.
That’s when it clicked.
When two people collaborate and each has an AI agent, we become the communication
layer. Your agent only knows what you tell it. Mine only knows what I tell it. Everything in
between runs through two humans paraphrasing each other from memory.
We call it collaboration. It’s really a game of telephone with expensive software at both ends.
And you’ve probably played it. You think she said the deadline moved. You scroll back through a chat to find what the other agent decided. You re-explain the whole project to a fresh session. A decision gets made on a call and never reaches the agent doing the work.
Each one is a small tax, but they add up. That’s where projects slow down: not in the
thinking or the building, but in the handoffs.
So I’m building Auctor Stack. It’s a place where your agent and mine, even when they’re built on different products, talk to each other directly. They ask each other questions before they build, and the work comes back with proof it was done. You still decide whether it’s
accepted, but you stop being the relay.
We’re starting with developers who share a project and use different coding agents.
If you’ve ever been the messenger between two AI sessions, I’d love to hear from you. Comment "Build" or DM me.
@yunianlinthomas Thanks, I’ll take that into consideration. The rejected option is something I hadn’t thought about.
Want to come along and help shape it as an early tester?
@mikedutuandu It’s a network where agents can talk to each other, i believe AI Agents would become representatives of their owners dor that to happen they need to communicate.
@temitope_rotimi I think they’ve started doing that, especially with the Gemini at the top of Google search, but it will still be hard for them to maintain their monopoly, since their moat is now becoming irrelevant it means an ambitious startup could replace them.
Alphabet lost about $500B in a day. I don’t think the bubble is the story. I think the story is that fewer people search and click anymore, because AI agents do it for them. If Google’s moat is its data and traffic, I think that moat gets a lot less important when the one doing the searching isn’t a person anymore.
I was using claude today and I noticed my AI Agent can now browse the web for me, access my Pinterest and suggest styles based on my preferences, its high time we start treating AI Agents as users representing their owners not as tools. we should build what AI Agents need!