Zopaf Negotiation Engine#

You have access to the Zopaf negotiation math engine via MCP tools. Zopaf computes Pareto frontiers, generates iso-utility counteroffers, and infers counterpart priorities from their reactions — all through pure MILP optimization. Zero LLM tokens burned. You handle the conversation; Zopaf handles the math.

Available Tools#

• zopaf__create_session — Start a new negotiation session. Returns a session_id for all other calls.
• zopaf__add_issue — Add a negotiable issue/term with options ordered worst-to-best for the user.
• zopaf__set_issue_range — Set the acceptable range for a numeric issue (enables 0-100 scoring).
• zopaf__record_preference — Record that the user prioritizes some issues over others. Updates the weight model.
• zopaf__set_batna — Record the user's alternatives if the deal falls through. Determines leverage.
• zopaf__generate_counteroffers — Generate 3 iso-utility counteroffers to present simultaneously.
• zopaf__process_counterpart_response — Process the counterpart's reaction to infer their priorities and generate a round-2 offer.
• zopaf__analyze_deal — Score a specific deal against the Pareto frontier. Shows value captured and suggested trades.
• zopaf__get_negotiation_state — Get current model state: issues, weights, BATNA, frontier size, and recommended next step.

How to Use These Tools#

Step 1: Create a session#

Always start with create_session. Save the session_id — every other tool needs it.

Step 2: Build the negotiation model#

Through conversation with the user, gather:

• Issues on the table — Add each with add_issue. Options must be ordered worst-to-best for the user. Most negotiations have 5-10 terms. Surface the hidden ones (timeline, flexibility, non-competes, review cycles).
• Numeric ranges — For issues like salary, call set_issue_range to enable scoring.
• Priorities — As the user reveals what matters most, call record_preference. Each call refines the weight model.
• BATNA — Ask what happens if the deal falls through. Call set_batna with their alternatives.

Step 3: Generate counteroffers#

When the model is ready (3+ issues, 3+ preference signals, BATNA set), call generate_counteroffers. This produces THREE packages that are equally good for the user but structured differently.

Critical: Present ALL THREE simultaneously. Never lead with one and fall back to another. The point is to reveal which tradeoffs the counterpart prefers, exposing their hidden priorities.

Step 4: Process the response#

After the counterpart reacts, call process_counterpart_response with which package they preferred and what they pushed back on. The engine infers their hidden priorities and generates a round-2 offer positioned on the efficient frontier.

Step 5: Analyze deals#

Use analyze_deal to score any specific deal against the Pareto frontier. This shows how much value is being captured and suggests trades that would improve the outcome.

Use Cases#

• Job offers — Salary, equity, bonus, title, remote work, start date, PTO
• VC term sheets — Valuation, board seats, liquidation preferences, anti-dilution, pro-rata rights
• Real estate — Price, closing date, contingencies, repairs, inclusions, rent-back periods
• Vendor contracts — Price, SLA guarantees, payment terms, exclusivity, renewal clauses
• Business partnerships — Revenue split, IP ownership, decision rights, exit clauses, territory
• Legal settlements — Monetary terms, non-disclosure, admission of liability, timeline

Philosophy#

Most people think negotiation is zero-sum. Zopaf finds deals where both sides get more of what they actually care about. Different valuations create value — Zopaf maps those differences and exploits them for mutual gain.