Sarah Prepares for the Negotiation

Dear Negotiation Explorer,

Welcome to issue 32 of our NegoAI series.

Last week, Sarah built her AI negotiation system. Six folders. A knowledge base extracted from her own documents in twenty minutes. The foundation was in place.

But a foundation isn't a strategy.

This week, Sarah uses the system. Not in theory — step by step, the morning before her biggest negotiation of the year. One scenario. Every step. From opening the laptop to walking into the room.

Sarah's negotiation with the consulting firm is at 2 PM. Seven hours.

She's prepared before, of course. Spreadsheets, internal memos, a few bullet points on her phone. But this time, she has something different: a system that already knows how she thinks.

She opens Claude. Her project is set up from last week — the Knowledge Base is already loaded. She doesn't need to explain her negotiation philosophy, her approach to anchoring, her views on relationship preservation. The AI knows all of that. It's in the KB.

Now she adds the deal-specific context.

Sarah attaches four things:

1. Her Knowledge Base — already in the project from last week. Eight areas of her negotiation expertise. Her philosophy, her frameworks, her industry knowledge. This is the document that makes every response personal.

2. The counterpart memory — a document she started building after the last contract review, six months ago. How this firm negotiates. Their partner who leads renewals — aggressive opener, softens in round two, sensitive about the firm's reputation. The procurement lead on their side who quietly signals flexibility before the partner does. Patterns she noticed. Surprises from last time.

3. The case materials — the internal performance review showing underdelivery. The competitive bids from three rival firms. The current contract terms. The analysis her team ran showing that AI tools now replicate 40% of the firm's routine strategic work.

4. The deal brief — a structured summary she wrote last Friday. What's being negotiated (renewal of $18.5M consulting contract). The key parties. Her interests and priorities. The timeline (contract expires in 90 days). What she needs from this meeting (agreement on restructured terms, or a clear signal that they're not going to move).

Four documents. Three minutes of drag-and-drop. The AI now has everything it needs — not just the facts of this deal, but the way Sarah thinks about deals.

This is the shift we discussed in issue 5. The prompt is maybe 20% of what determines output quality. The other 80% is the context you bring. Sarah just brought all of it.

Here's what Sarah runs:

If you've been following this series, you'll recognize the DNA.

The structure comes from issue 2 — same five elements. Context (the attached documents). Specific outcome (seven numbered sections). Purpose (the "why I need this" paragraph). Standards (the quality criteria). The only thing missing is persona — Sarah doesn't need it here because her KB already defines the lens.

But look at what's different from issue 2's metaprompt.

Issue 2 asked for "my interest ranking" and "the other party's BATNA" — standalone items. Sarah's prompt asks for a complete preparation workflow: know yourself, know them, map the landscape, expand the pie, plan the approach, anticipate their moves, distill for action. Seven sections that build on each other.

And the quality criteria don't just say "be thorough." They say: use my Knowledge Base, use the counterpart memory. They point the AI at the specific context Sarah loaded. Without those documents, this prompt produces generic output. With them, every section is grounded in how Sarah actually negotiates and what she actually knows about this firm.

The prompt doesn't do the heavy lifting. The context does.

Sarah hits enter. Three minutes later, she has a complete preparation package.

Her interests, ranked. Not a generic list — a ranking that reflects her KB's emphasis on relationship preservation alongside value capture. The AI flags that her deal brief mentions "maintaining the relationship" but her KB shows she's historically willing to walk away when performance gaps are documented. It asks her to clarify which takes priority here. That's not generic advice. That's the system catching an inconsistency she didn't notice.

The counterparty analysis, informed by memory. The AI doesn't guess that the partner will anchor aggressively — it reads it from the counterpart memory. It doesn't assume the procurement lead is the decision maker — it knows from Sarah's notes that the partner overrides on anything above 10% movement. It builds the profile from intelligence, not assumptions.

Creative options she hadn't considered. Because the AI knows Sarah's industry (pharma procurement) and her approach to value creation (from the KB), it doesn't suggest generic "bundle services" ideas. It suggests restructuring the engagement model — replacing the firm's junior consultants on routine work with Sarah's internal AI tools, while concentrating the firm's senior partners on the complex strategic work they were originally contracted for. That's not something a blank ChatGPT session would produce. It's the intersection of Sarah's case materials (the AI tools analysis) and her KB (her approach to creating mutual gains).

Scenario planning grounded in behavior. One scenario: the partner opens aggressively, citing the firm's "strategic value" and the risk of switching. Another: the partner comes in conciliatory, preemptively offering a discount to avoid losing the account. A third — the one Sarah wouldn't have expected: the partner proposes a completely different engagement model, pivoting from consulting to a hybrid advisory-plus-technology offering. The AI flags this as lower confidence but worth preparing for, given industry trends in the counterpart memory.

A one-page session brief. The numbers. The arguments. The decision points. What to listen for in the first five minutes that tells Sarah which scenario is unfolding.

Sarah reads through the output. She adjusts two things — her walk-away threshold and one creative option she knows won't fly internally. She saves the output to her Outputs folder. She prints the session brief.

7:45 AM. Thirty minutes, start to finish.

Sarah's prompt is specific to her deal. Yours should be too.

The structure stays the same — seven sections, quality criteria, the "why I need this" opening. What changes is the specifics.

Replace Sarah's deal details with yours. The contract value, the timeline, the leverage points, the counterparty dynamics. Write your own "why" paragraph — what decision does this analysis support? What do you need to walk in with?

The context stack stays the same: your Knowledge Base, your counterpart memory (even rough notes count), your case materials, your deal brief. If you built the system from last week's exercise, you already have most of this.

The prompt works because the system works. The seven sections are the framework. Your context is what makes it personal.

Let's step back.

Eight issues ago, Sarah — like most professionals — used AI by opening a blank chat and typing a question. The output was generic. The effort was wasted. Every session started from zero.

Now:

Thirty minutes of preparation that would have taken a full morning without the system. And the quality isn't just faster — it's better, because the AI is working with her expertise, not around it.

That's the complete workflow. Not eight separate techniques — one system, working together.

Run the preparation prompt on a real negotiation.

Step 1 (5 minutes): Pick your most pressing upcoming negotiation. Write a deal brief — what's being negotiated, who's involved, what you need from the next interaction.

Step 2 (5 minutes): Gather your context. Your Knowledge Base (from last week's exercise). Any notes on the counterparty. The deal materials.

Step 3 (5 minutes): Adapt the prompt above. Replace Sarah's details with yours. Write your own "why I need this" paragraph.

Step 4 (15 minutes): Run it. Read the output. Adjust what needs adjusting. Save the output to your system.

Then ask yourself: is this better than what you'd get from a blank chat? If the answer is yes — and it will be — you've felt the difference between a prompt and a system.

Sarah's preparation prompt gave her a strategy, scenarios, and a session brief. But there was something else running in parallel — something she hadn't mentioned yet.

While the preparation prompt analyzed the deal, a second AI agent was analyzing the person. The consulting firm's lead partner. His LinkedIn profile. His public talks. His writing.

The output: a full behavioral profile. DISC insights. Thomas-Kilmann conflict style. Specific clues on how to build trust, how to communicate, and how to negotiate with this particular person.

Next week: Sarah's secret weapon. The Kahneman behavioral agent — and what it told her about the person sitting across the table.

Questions? Reply directly — I read every response.