The question
AI can play more and more of the instruments. It can’t decide what the music is for. That part is yours. This is the question underneath everything InsightsOut does, and this page is where it lives.
Why this question
An AI system may be able to produce an answer, a recommendation, even a decision. That doesn’t settle whether it should, who looks at it afterward, or who’s accountable if it’s wrong. Most teams haven’t asked. They find the line after something fails, instead of drawing it on purpose.
The line between human and agent is drawn, not found. What stays human, who decides, and who answers when it goes wrong. Those are leadership questions before they are technical ones, and they belong to the people doing the work.
Everyone’s answer is different. This isn’t a debate about whether AI is good or bad. It’s a practice: look honestly at what’s changing, remember what has carried people through change before, and choose.
How to decide
If the honest answers are nobody, yes, yes, and no, the work stays human for now. Not because the tool is bad. Because nobody is standing behind it.
Then sort the work
For each part of a real workflow, pick one. Then test it with real work, notice where it fails, and redraw it. The boundary is a practice, not a document.
Where the question lives
A small-group conversation in San Francisco about what we want to protect, practice, and choose. Worried or excited, experienced or new to AI, you belong in the room. You leave with one choice made on purpose.
Saturday, October 10 · 2:00 PM · SF Commons, 540 Laguna St · Free for members, $10 suggested for othersNext date on Luma
When Agents Join the Team. The same question asked about one team’s real workflows: what stays human, what the agents own, where review is required, and who answers for each. You leave with a one-page Human-Agent Team Agreement. In public, and inside companies.
How do people keep agency, judgment, trust, and dignity as AI changes roles, teams, and decisions? We ask three questions when people register for a session and publish what holds up, with the sample, the version, and the limits. Aggregated themes only, with consent.
The essay this page is built on, and the newsletter of the same name: field notes from the rooms, patterns from the research, and the questions still open. A few times a month.
Newsletter
Field notes, research, and upcoming conversations about AI, work, leadership, and human agency. A few times a month. Registering for an event never adds you to this list.
Who’s asking
A decade building the systems, at EY, Neo4j, and a CTO seat where his team put LLM workflows into a product. Now an ICF-certified coach working on the people side of the same change, and navigating it himself. He won’t tell you where your line should be.