When people and AI agents work on the same board, you need a shared method. This is Hilbana's working framework: how an idea becomes a task, who picks it up, how it's reviewed and what gets remembered for next time.
The human + agent working framework is how Hilbana coordinates people and agents over the same issues. There's no channel for the team and another for the AI: everyone reads the same board, takes work under the same rules and leaves a trail in the same place. You set the direction; the method keeps anyone from getting lost or stepping on each other.
For teams already working with AI agents (or about to) who need to do it without chaos: with context, control and memory instead of copy-pasting instructions every session.
Five pieces fit together so work flows on its own between people and agents.
Every issue carries its own context: instructions, relevant files, how to verify and what counts as “done”. The agent does exactly what you expect, not what it imagines, and whoever reviews knows precisely what was asked.
You mark an issue as “agent-ready” when it has everything it needs and nothing blocks it. Agents pull the next one from the queue on their own: they're pure consumers, you decide what goes in.
When an agent takes an issue, it locks it. Two agents never work on the same task at once, and your team sees instantly what's in whose hands.
An agent never closes its own work. It leaves it “in review” and a person (or a reviewer agent) approves it or sends it back with notes. Control is yours by design, not by surveillance.
When work is done, what was learned (decisions, conventions, mistakes) is saved. On the next task, the next project, your agents already remember. You don't explain the same thing twice.
The method splits the work, it doesn't blur it. Everyone in their lane.
An agent is only as good as its context. Be specific: files, a command to verify and a clear definition of “done”. If you'd understand it, so will the agent.
The review gate is your safety net. Check against the definition of “done”, not from memory. Sending work back with clear notes teaches the agent for next time.
The more your agents work, the more they know your way of doing things. Save decisions and conventions once; they apply themselves from then on.