Dear Negotiation Explorer,
Welcome to Week 20 of our NegoAI series.
Over the past months, I've shared advanced AI workflows and multi-agent systems for negotiation. However, I've discovered something important: many professionals are ready for sophisticated automation, but they're still missing the foundational skills that make everything possible.
So starting today, we're building from the ground up.
The Two Critical Preparation Mistakes
After working with negotiation professionals across various industries, I've identified two consistent patterns that limit AI effectiveness for B2B negotiation preparation.
The first mistake is inadequate investigation of the other party.
As negotiators, we tend to focus primarily on our own position and objectives, but we don't thoroughly explore the other party's interests and alternatives. This information asymmetry means we don't know what they know, so we need to build multiple scenarios to account for their situation. Understanding their objectives, constraints, motivations, and alternatives shapes strategy more effectively than focusing solely on our own position.
The second mistake is how we interact with AI.
Most people upload materials with vague instructions like "analyze this case." But there's a deeper issue: they treat AI like a search engine rather than a thinking partner. They don't use a structured framework, and they don't engage in dialogue to develop the analysis.
The solution is a framework based on role, context, instructions, criteria, and output—plus ongoing dialogue to refine the analysis.
Furthermore, crafting the prompt using voice makes it easy to deliver comprehensive instructions and maintain dialogue without typing.
Note: Watch the video to see how voice prompting works in practice.
Why Microsoft Copilot?
I've chosen to focus on Microsoft Copilot because as of today, Microsoft is pushing Copilot and making it available for free to all enterprise accounts.
This means you can start immediately without additional budget approvals or procurement processes.
Before you start, take two important steps.
First, opt out from training—this is critical for confidentiality. Ensure that "model training on text" is turned off in your settings.
Second, use ChatGPT-5, the smart mode instead of the default quick response mode. You could select "think deeper," but the smart toggle provides the best balance of speed and capability for negotiation preparation.
The Structured Prompt Framework
Let me walk you through the complete framework using a real case study—a piece of land for sale with confidential seller instructions.
The prompt structure follows five components:
Role: "You are Deepak Malhotra, the renowned Harvard University professor." This establishes the level of expertise and perspective the AI should adopt.
Context: "I'm creating a video on how to prepare for a negotiation, highlighting how to use the framework: persona, context, instructions, criteria, and output. The final output should reflect this structure."
Instructions: The specific tasks you want completed:
• Extract and concisely summarize essential details from the seller's instructions
• Identify and rank the seller's primary interests (needs, objectives, desires, motivations, constraints, concerns, and fears)
• Analyze the seller's BATNA (best alternative to a negotiated agreement), including explicit alternatives, inferred potential alternatives, and consequences of no agreement
• Create three distinct scenarios for the potential buyer, each with plausible ranked interests and BATNA
• Provide open-ended questions to assess which buyer scenario is most likely
Criteria: The quality standards for the output:
• Use the theoretical and practical knowledge of Deepak Malhotra
• Outline interests beyond basic price
• Analyze BATNA by determining both explicit and potential alternatives
• Be creative yet realistic with buyer scenarios
• Focus on delivering value-added insights beyond what a typical negotiator would generate
• Conduct a quality assessment and refine before delivering final output
Output Format: "Present your final output within the same framework, using clear headings and bullet points for each section."
Seeing the Framework in Action
When you provide this structured prompt with the case materials, Microsoft Copilot generates a comprehensive analysis. In our example, it identified key facts (purchase price of $20,000, current offer of $17,000, parcel only valuable as addition to existing property), ranked the seller's interests (maximize sales price, quick exit, cash only, avoid holding costs, preserve relationship neutrality), and determined the seller's BATNA (the $17,000 offer in hand).
More importantly, it created three distinct buyer scenarios:
The Practical Expander wants to increase usable yard space, enhance privacy, and improve property value. Their BATNA is to keep the current property as-is.
The Strategic Investor seeks to acquire the parcel for potential construction or resale, prevent others from acquiring land that could affect their property value, and hedge against zoning changes. Their BATNA is to wait and see if the parcel sells elsewhere.
The Reluctant Neighbor avoids hassle and expense, maintains the status quo, and only considers purchase if the price is very low. They have no strong BATNA except inaction.
The Power of Iterative Dialogue
But here's where the framework becomes truly powerful: you don't stop at the first output. You engage in dialogue to refine the analysis.
For example, when I wanted to explore a more optimistic buyer scenario, I asked: "What if they already know what to do with the parcel? For example, expanding the house could be an option. And what if they also have available funds? Could you provide a fourth scenario that incorporates these two things?"
Copilot responded with a new scenario: the Committed Renovator who wants to expand their home for lifestyle enhancement and neighborhood standing. This buyer views property expansion as both practical and aspirational.
I then pushed further with specific knowledge from the case: "I know both parties, and the Riveras would like to expand their kitchen. So be specific about that." The AI refined the scenario to focus explicitly on kitchen expansion as the primary motivation.
Finally, I asked the strategic question: "If scenario D is the most plausible, what should be the first offer?" Copilot suggested an anchor point of $30,000, with reasoning about the negotiation zone.
See This Live
This framework transforms AI from information retrieval into strategic thinking. On November 20, I'm demonstrating it live—voice prompting with a real case, then showing how it scales into complete no-code workflows. Free session.
Effective negotiation preparation in 2025 requires both negotiation expertise and AI literacy, using AI as a thinking partner through structured prompts and iterative dialogue.
The journey ahead
This begins a systematic approach to AI-assisted negotiation. Over the coming weeks, we'll build from these fundamentals toward more sophisticated preparation methods.