DLDesheng LuContact
All work
MVP / UAT-readyFuture-state MVPAnonymized

AI-Native Negotiation / Flagship product

Negotiation Pro: Zero‑to‑Close

One governed negotiation workspace turns commercial evidence into position, strategy, rehearsal, and human-controlled live support, then carries the agreement into reporting and reusable learning.

My mandate

Define and lead the UAT-ready Zero-to-Close MVP, connect the working practice experience to real case evidence, and establish the quality, authority, audit, and human-approval gates required for assisted negotiation.

Simulated negotiation workspaceSimulated product scenario
Product viewSimulated negotiation workspace
Simulated Zero-to-Close AI-native negotiation product showing case evidence, strategy, practice, live support, and outcome review

Public-safe reconstruction using fictional or anonymized information. It demonstrates product mechanics, not an original product screen.

Portfolio product state
MVP target · UAT-ready Zero-to-Close system
Evidence basis
Working negotiation-practice prototype completed; the assisted Zero-to-Close workflow remains under validation.
Primary users
Procurement buyers / Category managers / Sourcing leaders
Leadership scope
Strategy / roadmap / validation / cross-functional delivery

Evidence first

What is proven, what I changed, and what remains open.

A 90-second evidence map for hiring review. Synthetic product views explain mechanics; they never count as outcome proof.

Verified now
A working AI negotiation-practice and coaching experience, UAT readiness, and a documented layered evaluation methodology.
Next evidence gate
Verified UAT synthesis and expert quality evidence before extending from practice into real-case or live decision support.
01

User / workflow evidence

Reported · receipt pending

UAT began and subsequent feedback was reported as broadly positive, but respondent coverage, positive and negative themes, edge-case trend, and resulting priority decisions are not yet connected to a public-safe synthesis.

02

Product decision

Verified record

I kept the verified core focused on realistic practice, coaching, feedback, and improvement over time, while treating the broader Zero-to-Close workspace as a staged future-state product rather than a completed live-negotiation capability.

03

AI / system boundary

Supported synthesis

AI powers scenario personas, practice interaction, coaching, and feedback; product controls preserve case state and evaluation; the negotiator retains authority. Real-deal preparation, live support, and autonomous negotiation are not current verified capabilities.

04

Evaluation → change

Verified record

The evaluation method includes deterministic tests, real-session regression cases, designed behavioral scenarios, automated judges, repeated sampling, baseline comparison, and staging checks. Candidate pass results are not yet available.

05

Outcome / boundary

Supported synthesis

Phase 1 records support a core simulation, scenario flow, user-facing practice experience, initial coaching direction, and controlled UAT approach. Adoption, current release clearance, commercial outcome, and live assistance remain unproven.

AI product decision record

The choices behind the product, architecture, and release gate.

Use an LLM for realistic rehearsal and contextual coaching; keep commercial authority and case truth outside the model.

01

AI fit

Verified record

Use an LLM for counterpart simulation, conversational practice, coaching, and feedback across repeated attempts.

Negotiation quality depends on adaptive language, behavior, and context that fixed scripts cannot reproduce well.

Alternative consideredA scripted branching simulator is easier to score but too predictable to build transferable judgment.

02

Knowledge & context

Supported synthesis

Ground recommendations in the persistent case packet, approved evidence, objectives, limits, and round history.

Commercial advice must explain which case fact or explicit assumption supports it.

Alternative consideredOpen-web RAG or model memory by default would mix generic negotiation advice with unverified case facts.

03

System architecture

Supported synthesis

Use one case-state workflow spanning practice, feedback, and learning; preserve human control over every commitment.

Continuity and authority matter more than creating separate agents for strategist, coach, and negotiator.

Alternative consideredLive action, MCP-connected enterprise tools, or autonomous negotiation remain gated until permissions, reversibility, and expert evals are proven.

04

Evaluation & release

Verified record

Combine deterministic regression, behavioral scenarios, repeated sampling, automated judges, baseline comparison, and expert review.

One conversation cannot reveal persona drift, unsupported leverage, authority violations, or unstable coaching quality.

Alternative consideredA single average score or user satisfaction signal would hide high-severity commercial failures.

Evidence I still need from the private project record3 open items
  1. 01

    UAT cohort, use cases, positive and negative themes, and the backlog changes they caused.

  2. 02

    Current build, test, deployment, and evaluation-pass receipts for the release candidate.

  3. 03

    One public-safe real-session example showing practice behavior before, feedback, and observed improvement after.

Verified = direct approved recordSupported = defensible synthesisReported = source receipt pendingDesigned = future test or control

01 / Executive impact

Why this mattered

Negotiation value disappears when evidence, preparation, live rounds, and the close live in separate tools.

Buyers repeatedly rebuild context across research, strategy documents, training, meeting notes, and reporting. Generic practice may improve confidence, but it rarely stays with the real case or preserves what the organization should learn after the deal.

My impact

I expanded a working practice experience into a governed Zero-to-Close product system.

Verified now: the current product core is an AI-assisted negotiation practice and coaching experience with controlled UAT readiness and a documented layered evaluation method. UAT synthesis, candidate pass results, real-deal preparation, live assistance, Zero-to-Close delivery, production adoption, and realized commercial impact are not claimed.

  1. 01
    Persistent context

    Make the real case the system of record

    I connected supplier, market, contract, spend, objectives, constraints, and round history across preparation, practice, live support, and close.

    One case from evidence to agreement
  2. 02
    Product priority

    Choose intelligence over theatre

    I prioritized evidence quality, strategy, credible counterpart behaviour, coaching, and decision guardrails before avatar-led presentation features.

    Value before visual spectacle
  3. 03
    Authority model

    Stage assistance behind evidence

    I defined explicit mandates, approval limits, auditability, escalation, and expert evaluation before the product can influence live commercial action.

    Human control at every commitment
  4. 04
    Learning loop

    Make every close improve the next case

    I designed plans, concessions, outcomes, feedback, and lessons to compound across rounds and future negotiations.

    Transaction to organizational memory

02 / Future-state MVP system

The complete Zero-to-Close MVP, with human authority preserved at every commercial decision.

Strengthen human judgment first. Preserve case evidence across preparation and rehearsal, then expand into live support only when evaluation and governance justify it.

North-star experience

In the MVP product system, a buyer brings a real case and approved evidence. The product diagnoses leverage, builds strategy, rehearses the conversation, supports the human during each round, and produces a reviewable close record. The negotiator retains authority over every commercial commitment.

Zero-context product contract

The user, the system, and the decision are explicit.

A cold reader should be able to describe the product without relying on internal context, AI terminology, or the rest of this portfolio.

Primary user
A buyer or category manager preparing for and conducting a consequential supplier negotiation.
Trigger
A negotiation case opens with objectives, evidence, constraints, and an expected close.
Inputs
Supplier and market evidence, contract and spend context, history, objectives, walk-away limits, stakeholders, and approved authority.
MVP product action
Diagnose leverage and gaps, build strategy, coach the user, run practice, support live rounds, capture concessions, and structure the close.
MVP output
A persistent case workspace containing the plan, rehearsal feedback, live decision support, round history, and close report.
MVP decision enabled
What to ask, offer, hold, escalate, accept, or carry into the next round.

How the product works

From signal to accountable action.

  1. 01

    Gather

    Bring supplier, market, contract, spend, history, objectives, constraints, and approved evidence into one case.

  2. 02

    Diagnose

    Assess leverage, interests, risks, alternatives, dependencies, and information gaps.

  3. 03

    Strategize

    Build targets, walk-away positions, scenarios, concessions, questions, and a negotiation plan.

  4. 04

    Coach

    Translate the case into expert-informed guidance, capability gaps, and preparation priorities.

  5. 05

    Practice

    Rehearse against credible counterpart personas and scenarios with structured feedback.

  6. 06

    Assist

    Support the human negotiator with context, evidence, prompts, and decision guardrails during the live process.

  7. 07

    Debrief

    Compare plan, behaviour, concessions, outcome, and missed signals after each round.

  8. 08

    Conclude

    Capture agreed terms, unresolved actions, lessons, approvals, and an executive-ready close report.

Interactive product walkthrough

Follow one negotiation from first evidence to close.

This walkthrough demonstrates the UAT-ready Zero-to-Close MVP. The working practice prototype is verified evidence; assisted stages remain a public-safe future-state reconstruction until promoted by approved UAT evidence.

Negotiation Pro: Zero‑to‑CloseNegotiation decision workspace
MVP / UAT-ready walkthrough
Simulated product scenario
Public-safe simulation

Every organization, supplier, geography, value, date, confidence measure, and output shown here is simulated. This demonstrates product mechanics and decision design, not an employer interface or an actual project result.

Scenario

A fictional category team is preparing for a complex service renewal with incomplete leverage and evidence.

Decision owner
Category lead and negotiator
AI role
Ground, coach, and preserve control
Boundary
No autonomous commercial commitment
01 / Frame

Turn a commercial ask into a negotiation contract.

The product captures the objective, constraints, decision rights, and missing facts before recommending a posture.

Human gateMandate confirmed
Negotiation mandateClarification complete
Commercial decisionHow should the team respond to a fictional service provider's renewal proposal without trading away critical operating terms?
Objective
Protect service continuity and improve total value
Decision owner
Category lead
Non-negotiable
Human owner approves every commercial commitment
Unknown
Supplier flexibility across term and service levers

Product principleThe product improves the decision without replacing the accountable negotiator.

Use arrow keys to move between workflow stages

AI operating and evaluation model

The model reasons. The product controls. The human remains accountable.

This product view presents the complete MVP/UAT-ready Zero-to-Close system as a future-state public-safe reconstruction. The working practice prototype and assisted roadmap are verified evidence; live assistance and any bounded agent authority require approved expert and UAT evidence before promotion.

01Deterministic product

Case state, authority, and commercial guardrails

Targets, walk-away limits, approved concessions, document versions, round history, and required approvals remain structured and auditable.

02AI-assisted reasoning

Diagnosis, coaching, simulation, and synthesis

AI assembles evidence, identifies gaps, proposes strategy, simulates counterpart behavior, and compares the plan with each negotiation round.

03Accountable human

Mandate, judgment, and agreement

The negotiator owns objectives, evaluates advice, controls live action, approves concessions, and concludes the deal.

Representative failure mode

Fluent advice ignores case evidence, invents leverage, or recommends a concession outside the buyer's authority.

Designed product control

Evidence-linked recommendations, explicit assumptions, expert review, scenario comparison, authority limits, and human approval before consequential action.

Illustrative acceptance benchmark

every strategic recommendation cites case evidence or an explicit assumption and remains inside the approved mandate; unsupported leverage is rejected.

Decision record

How I led it, decision by decision.

The executive impact summary is substantiated here through the choices I made, the boundaries I held, and the next investment gate I defined.

Discover

Make the real case the system of context

Supplier, market, contract, spend, history, goals, and constraints should inform every stage from initial diagnosis through final reporting.

Prioritize

Prioritize decision intelligence before visual theatre

Evidence quality, strategy, credible counterpart behaviour, coaching, and guardrails create value sooner than avatars or presentation-heavy features.

Retain

Build a compounding negotiation memory

Plans, sessions, concessions, outcomes, skill signals, and lessons should improve the next round, the next negotiation, and eventually organizational capability.

Trust

Stage additional agent authority behind evidence

Agent advice and action need expert evaluation, explicit mandates, financial and policy limits, auditability, escalation, and human approval appropriate to the stakes.

MVP readiness and evidence

Complete product story. Explicit evidence basis. One promotion gate.

The case shows the intended MVP/UAT-ready product in full. Verified delivery evidence, future-state mechanics, and post-UAT value remain separate so the page can be promoted later without being rebuilt.

01Verified today

Evidence basis

  • Phase 1 records support a scenario-based negotiation practice simulation, user-facing session flow, initial coaching and feedback direction, and controlled UAT readiness.
  • The documented evaluation method includes deterministic checks, real-session regression cases, designed behavioral scenarios, automated judges, repeated sampling, baseline comparison, and staging checks; current candidate pass evidence is not available.
  • Reported UAT sentiment and edge-case improvement have not yet been connected to respondent coverage, a balanced synthesis, a defect trend, or the resulting priority decisions.
02Future-state MVP

MVP / UAT-ready product system

  • One governed case connecting evidence, diagnosis, strategy, coaching, practice, live support, debrief, and reporting.
  • Persistent negotiation memory across sessions, skill profiles, concessions, outcomes, and same-case comparison.
  • Human-controlled authority, expert evaluation, auditability, escalation, and explicit approval before commercial action.
03Post-UAT scale

Value and scale thesis

  • Reduce fragmented preparation and preserve decision context from first analysis through final close.
  • Make expert-informed coaching and realistic practice available inside the flow of a real negotiation.
  • Turn negotiation outcomes and lessons into reusable organizational intelligence rather than isolated individual experience.

Investment-grade product judgment

The gates required before expanding from practice to live decision support.

A polished prototype is only the beginning. The product lead must make value, risk, ownership, evidence, and the next investment decision legible.

Adoption wedge

Start with the real case and the work buyers already have to complete from preparation to close.

Quality gate

Evaluate evidence use, strategic advice, simulation realism, live-assist relevance, and reporting quality with experts.

Autonomy gate

Increase agent authority only when mandate, limits, approval, escalation, auditability, and observed quality support it.

Outcome model

Track preparation time, case completeness, strategy quality, practice improvement, negotiated outcome, compliance, and reusable learning.

This is a capability signal, not a claim that this public case carried a disclosed budget or realized a specific dollar value.

Evidence promotion gate

What converts this future-state MVP into a verified public result.

Approved expert and UAT evidence across advice, practice, live support, guardrails, and close reporting.

  1. 01

    Validate the assisted Zero-to-Close journey with real negotiation cases

  2. 02

    Evaluate advice, simulation, live-assist, and reporting quality with experts

  3. 03

    Define outcome, learning, trust, and guardrail metrics

Next flagshipProcurement AI Hub