One worked example
Multi-agent team meetings: how it works
You're the chair. The agents are the room. The disagreement is the point.
Most of the time you work with one agent at a time. But some decisions need more than one head in the room — the ones where the answer depends on who you ask. When a project stalls on a real fork in the road, you can call a meeting: pick the agents you need, put the question on the table, and let them argue it out in front of you. Here's one such meeting, start to finish.
The situation
You're three months into a nonfiction book — a history of a small river town that boomed and then emptied out. You've done the archive work in your Research space, drafted four chapters in your Build space, and you're stuck. Not on a sentence. On the shape of the whole thing.
Two chapters want to be a strict chronology — founding to collapse, decade by decade. But the most alive material you've found is a set of family stories that cut across the timeline, and they fall flat when you force them into order. You've rewritten the outline four times. Every version fixes one problem and breaks another.
This isn't a question one agent can answer cleanly, because the right answer depends on what you weigh most: the history, the prose, or the reader who has to make it to the end. So you stop rewriting the outline and convene the room.
The convening
You open the meeting and pick the room
From inside the project, you call a team meeting. You don't invite everyone — a crowded room is just noise. You pick three agents whose angle actually bears on the fork you're stuck at, and you state the question plainly. You are the chair: you set the agenda, you run the room, and you decide when it's done.
The dialogue
They each weigh in — and they disagree
You ask the question once. Each agent answers from its own discipline. They're not performing agreement for your comfort; they pull in different directions, because the disciplines genuinely do. That tension is exactly what you came for.
You · chair
I'm stuck on structure. Strict chronology keeps the history honest but kills the family stories. Threading the family stories through breaks the timeline. Which way should this book go?
Mara · research historian
Chronology, and don't flinch from it. The whole argument of this book is causal — the railroad came, then the mill, then the layoffs, then the exodus. Scramble the order and a reader can no longer see why the town emptied. Lose the sequence and you lose the history. The family stories are evidence; sort them into the decade they belong to.
Dev · developmental editor
I disagree, and I think Mara's confusing a timeline with a book. Readers don't feel a decade — they feel people. Your strongest pages are the family stories, and a chapter that opens “In 1931, the mill's output fell 40 percent” loses them by the second sentence. Lead with the people. Let the chronology be the spine you hang them on, not the thing on the page.
Ines · reader advocate
Both of you are arguing about the author's convenience, not the reader's experience. A reader doesn't need the history front-loaded or a parade of anecdotes — they need to know, early, why they should care that this town died. Neither a clean timeline nor a great story answers that on its own. If you make me choose between Mara's rigor and Dev's warmth, I'll put the book down before either one pays off.
You · chair
So chronology protects the argument, the family stories carry the feeling, and neither one tells the reader up front why it matters. What if the stakes don't live in the structure at all?
Ines · reader advocate
Then say them out loud. A short prologue — present day, the empty main street — that asks the question the whole book answers. Now I know why I'm reading.
Dev · developmental editor
That I can build on. Keep Mara's chronology as the chapter order, but open each chapter on a person, then pull back to the decade. The family stories aren't sidebars — they're the way into each chapter.
Mara · research historian
I can live with that, with one condition: the causal chain has to survive. If a family story sits in the wrong decade for drama, it breaks the argument and I'll flag it. Anchor each one to its real date and I've no objection.
Nobody “won.” You didn't get a vote or a tidy consensus — you got three honest positions, the friction between them, and a synthesis that none of the three would have reached alone. That synthesis is yours to ratify, because you're the one who has to write the book.
The decision
You keep what's useful — one artifact comes out
You don't adopt any single agent's answer wholesale. You take Ines's prologue, Dev's chapter shape, and Mara's non-negotiable on dates, and you fold them into one decision. The room surfaced the trade-offs; the call is yours. What lands back in your Build space isn't a transcript to re-read — it's a single, concrete structural plan you can write against tomorrow morning.
Structural plan — revised outline
- Prologue: present-day empty main street; pose the book's question.
- Chapters in chronological order (the causal spine — railroad → mill → layoffs → exodus).
- Each chapter opens on a person, then pulls back to the decade.
- Constraint: every family story anchored to its real date — no drama-driven reordering.
Decided by you. Saved to your Build space.
The meeting took twenty minutes and it broke a three-week stall. Not because the agents agreed — because they didn't, and you were in the room to make the call.
Theta Meridian is in private beta.
If you have a project stuck on a decision that needs more than one head in the room, request access.