AI agents are beginning to do work that used to belong to people. They draft, analyze, recommend, decide, and sometimes act. But most teams have not agreed on where the line is.
Who decides? Who checks the work? Who answers when an agent gets it wrong? What should always stay human?
Without a shared answer, everyone creates their own rules. One person lets the agent send the client email. Another rewrites everything. Someone else stops trusting the work, or hides how much they use AI. The managers end up refereeing every case by hand.
This is a working session for people building, leading, or working inside human-agent teams. We use real workflows to turn unspoken assumptions into a shared way of working.
What we’ll do
- Put the real work on the table. Three to five recurring workflows that matter, as they actually run, not the tidy version.
- Decide what should stay human. Anything the team wants to keep human has to pass four questions: who answers for it, who it affects, whether it needs judgment, and whether the people affected can push back. This is where fear and turf get separated from judgment.
- Define what agents can own. And what an agent may prepare but a person must decide.
- Set the points where human review is required. For everything an agent touches, one named person who owns the outcome.
- Agree how the team handles disclosure, escalation, feedback, and mistakes. Then draft the sentence the team would say to a customer, a new hire, or itself about where the line is.
What you leave with
- A one-page Human-Agent Team Agreement.
- Clearer decision rights and human boundaries.
- A simple review practice the team can repeat as the technology and the work change.
Who this is for
- Founders, managers, team leads, transformation leaders, and People leaders.
- People working on teams where AI agents are beginning to take on meaningful work.
- Heads of People or Operations who are being asked to write the AI guidelines and want them to come from the work, not from a template.
This is not an AI tool demonstration or a prompt-writing class. It is about how the team works when both people and agents are involved.
What to bring
One real workflow where people and AI agents already work together, or soon will.
Inside a company
The private version runs for a leadership team, or for one working team and its manager as a pilot. Half a day, in your office or online. The team sorts its real recurring work, names who answers for each part, and writes down the decisions that need to go up. From there, most teams keep it alive with a monthly manager practice and, when a team is ready, a bounded AI Enablement Lab around one workflow.
The details
- Format: Facilitated working session, whole team in the room
- Length: 2 hours in public · half a day inside a company
- Next public date: Thursday, October 29, 2026, 6:00 PM (register)on the Luma calendar
- Where: SF Commons, San Francisco · your office in the Bay Area · online
- Price: Public: free for SF Commons members, $10 suggested for others · Private: scoped on a call
Your facilitator
Nima Imani is an ICF-certified, ontologically trained coach with more than ten years in software, data and AI, including EY and Neo4j, and a CTO seat building LLM workflows. He runs the weekly founder coaching group at SF Commons in San Francisco and founded InsightsOut.