Insight: AI as Negotiation Partner — Role-Play the Deal Before It Happens

“I dropped it in an executive committee meeting last night, and it rolled out exactly as it was role-played by Claude. I’ll never tell any of the other people on the board how I actually did that.” — Don Back

Session context: 2026-06-04_Mastermind — Don Back described using Claude to prepare for a not-for-profit board negotiation that had been festering for three years. He loaded the governing Act, bylaws, a meeting transcript, and stakeholders’ positions — then ran a full back-and-forth negotiating rehearsal before the real meeting. The result matched the role-play.

Core Idea

Most people use AI to answer questions: “what does this clause mean?” or “what are the arguments for my position?” Don Back used it as a negotiating rehearsal room — feeding it not just the rules but the people: everyone’s position from the meeting transcript, the governance Act, the bylaws, what he wanted to achieve, and a draft strategy with what he was willing to trade away. Then he ran the negotiation before it happened.

The key moves, in sequence:

1. Load the full context. Governing documents (the Act, the bylaws), a transcript from the meeting where all parties stated their positions, and your desired outcome. This seeds Claude with the knowledge of the room.
2. Give it a negotiating role. Instruct it to argue from the position of the other parties — not just acknowledge their concerns but actively push back as they would.
3. Iterate until the strategy is solid. Don interrogated the logic: “analyze this, audit this, what do you think about this?” The goal wasn’t to have Claude tell him what to do — it was to debate until the reasoning was tight.
4. Build the briefing note with cannon fodder. Don embedded seven recommended actions in the brief, of which he only wanted one. The other six were negotiable — deliberate concessions he was willing to make to secure the one outcome he needed.

The result: the meeting “rolled out exactly as it was role-played.” This isn’t because Claude predicted the future. It’s because running the rehearsal with full context forced Don to stress-test every position before walking into the room. He arrived knowing every objection, every counter-move, and what he was willing to give up.

What makes this different from normal research: Standard AI use for negotiation is informational — research the other side, understand the rules, know your BATNA. Don’s pattern is experiential: he ran the adversarial process inside a conversation before the real adversarial process happened. The friction of the back-and-forth is what clarified his thinking. Claude “wasn’t telling me the truth — but it was enabling me to think it through and get better and better at honing what I wanted to achieve.”

Why it works: The AI has been seeded with the stakeholders’ actual positions. It can simulate how they’ll respond more accurately than your imagination can, because it’s starting from their stated logic rather than your assumptions about it.

Practical Application

For any significant negotiation, board interaction, or high-stakes meeting where stakeholders have stated positions:

1. Record or transcribe the prior meeting — even notes are enough. Capture everyone’s stated position.
2. Load the governing documents — contracts, bylaws, regulations, whatever defines the formal constraints.
3. State your objective and your strategy to Claude. Be explicit: “I want X. I’m willing to give up Y and Z to get it.”
4. Ask it to role-play as the opposition — specifically to push back and not yield without data.
5. Debate until your reasoning holds — audit every position, test every counter-argument, refine the approach.
6. Produce the briefing note with deliberate structure: your real goal embedded among acceptable-to-trade positions.

Works equally well for: client contract negotiations, salary discussions, partnership terms, board votes, policy advocacy. Any situation where you enter a room with a goal and leave with a negotiated outcome.

Related Insights

• Insight - Tell It What to Do, Don’t Ask It Questions — The Posture That Makes AI Think — the delegation posture is what makes this work: “here’s my case, here’s the context, role-play this negotiation” vs. “what should I do?”
• Insight - Multi-Model Debate as a Decision-Making Accelerator — the debate structure here is applied to a real-world negotiation, not just an internal AI deliberation
• Insight - Multi-Model Debate as a Quality Control System for High-Stakes Work — same quality-through-adversarial-friction principle; this is the single-model version for external negotiations
• Insight - The Skeptic Command - Stress-Testing AI Answers Before You Act on Them — this is the Skeptic pattern applied to strategy rather than content
• Insight - Always Audit Your Plan Before You Build — The 18-Problem Discovery — audit-before-execute discipline; this applies it to interpersonal strategy
• Insight - Ask AI to Reverse-Engineer Your Conversation to Recover Hidden Frameworks — a related thinking-partner move: AI as a mirror to expose the structure of your reasoning

Evolution Across Sessions

This establishes the baseline for the Negotiation Rehearsal Pattern as a named technique. Prior sessions have covered debate structures (Insight - Multi-Model Debate as a Decision-Making Accelerator, 2025-09-25) and thinking-partner postures (Insight - Tell It What to Do, Don’t Ask It Questions — The Posture That Makes AI Think, 2026-05-28). New development: this session gives the first concrete, real-world account of using Claude to role-play a full multi-stakeholder negotiation before it happens — with the outcome confirming the rehearsal’s fidelity. The cannon-fodder tactic (embedding deliberate concessions in the briefing note) is a named strategy that belongs to this pattern.