AI · Multi-agent · '26
AutopilotHQ
An AI company in a box: nine specialist agent roles that research, create, launch and report.
- 9 agent roles
- 5 drafted
01
Nine defined roles across research, creative, ads, email and analytics; five are drafted with prompts and tool specs.
02
Agents run in parallel and hand work to each other, so throughput is not capped by the user.
03
The user stays in command: agents propose, act within limits and report back.
- Next.js
- Firebase
- LLM agents
The story behind it
Why it exists
Multi-agent coordination is the hardest open problem in applied AI.
Problem
One chat window handles one thread at a time. As soon as a task needs research, then creative, then a launch and a report, the user becomes the manager again.
Idea
Assign the work to a team of agents instead of a chatbot.