Enterprise AI Product & Portfolio Leader

Unilever | Horizon 3 Lab, Toronto

I turn complex enterprise workflows into trusted AI products.

I shape AI portfolio strategy, turn ambiguous workflows into governed products, and align cross-functional teams from MVP through evaluation toward enterprise handoff.

Senior AI product leadership

AI product strategy, made executable.

I shape portfolio choices, align cross-functional teams, and build the evaluation and governance path from MVP toward enterprise handoff.

Portfolio strategy

Shape business cases, roadmap priorities, and reusable capability bets.

Zero-to-one AI products

Advance platforms from MVP through evaluation toward enterprise handoff.

Matrix leadership

Align business, product, engineering, data, and AI around evidence and accountability.

Desheng Lu, Enterprise AI Product and Portfolio Leader
~€4M

Enterprise AI portfolio scope

3 AI platforms

Advanced from 0-to-1 MVP toward enterprise handoff

4 PMs + 20 engineers

Current AI portfolio matrix leadership

80K+ users

Shipped B2B platform across 3,000+ partner organizations

Selected product systems

Three AI product systems, built for accountable action.

Each flagship exposes the full product logic: the workflow being changed, where AI assists, where humans decide, and what evidence is required to advance from MVP into enterprise validation.
Enterprise AI orchestrationMVP · Structured evaluation / UAT

Procurement AI Hub

A category lead starts with a business decision, not an agent name. The Hub carries enterprise context through a reviewable 1-to-N specialist plan and returns one source-backed decision packet the owner can approve, revise, or escalate. I defined the orchestration model, capability contracts, shared controls, evaluation gates, and enterprise handoff path.

Evidence-backed maturityAnonymized

What I shaped

The operating model behind the interface.

  1. 01
    Orchestration model

    One governed entry point designed for an expanding specialist portfolio

  2. 02
    Control plane

    Reviewable 1-to-N plans, capability contracts, human authorization, and visible recovery

  3. 03
    Productization

    Coverage, evidence, evaluation, ownership, and enterprise handoff designed as one system

Workflow, decisions, and evidenceView product
Geospatial supply-chain intelligenceMVP / UAT-ready

Tier‑N Geospatial Risk Navigator

An event-driven geospatial intelligence product that traces a global disruption through potential hidden-tier and material exposure, then routes qualified evidence to an accountable owner. I converted Phase 1 feasibility into a UAT-ready product system with explicit usefulness, evidence-quality, human-review, and scale gates.

Future-state MVPAnonymized

What I shaped

The product decisions behind the risk workflow.

  1. 01
    Decision model

    Event-to-exposure model spanning location, hidden tiers, and material relevance

  2. 02
    Evidence quality

    Evidence states separating sourced facts, inference, and coverage gaps

  3. 03
    Accountability

    Human validation and an owned response before escalation

Geospatial workflow and evidenceView product
AI-native negotiation systemMVP / UAT-ready

Negotiation Pro: Zero‑to‑Close

An AI-native workspace that carries a negotiation from commercial evidence and strategy through rehearsal, human-controlled live support, agreement capture, and reusable learning. I expanded a working practice prototype into a UAT-ready Zero-to-Close product system, defining the quality, authority, audit, and human-approval gates required before live support.

Future-state MVPAnonymized

What I shaped

The operating choices behind Zero-to-Close.

  1. 01
    Workflow

    One persistent case from commercial evidence through agreement

  2. 02
    Authority

    Human-controlled live support with explicit authority gates

  3. 03
    Learning

    Close record and reusable learning without autonomous commitment

The complete zero-to-close systemView product

Product leadership system

Move the decision forward, not just the demo.

My product layer sits between AI capability and business adoption. I make uncertainty explicit, create decision structure, and protect the difference between technical promise and business proof.
  1. 01

    Frame

    Define the workflow, user decision, value hypothesis, and non-goals.

  2. 02

    Prioritize

    Balance value, capacity, risk, reuse, ownership, and timing.

  3. 03

    Evaluate

    Set quality bars, evidence, failure modes, human review, and UAT.

  4. 04

    Productize

    Design adoption, governance, ownership, handoff, and scale.

Earlier ventures and builds

Range, without losing the narrative.

Supporting work shows zero-to-one startup execution, scaled enterprise platforms, vertical SaaS thinking, and AI-assisted prototyping.

Startup product · earlier venture

TruWorld

A mobile product combining real-world exploration, local discovery, social mechanics, and partner-based rewards.
  • 45K+ community
  • 30+ retail partnerships
  • ~$500K completed strategic investment
View selected proof

Enterprise platform · shipped

Vendor Operations Platform

A unified B2B workflow connecting logistics reporting, partner support, visibility, and operational decisions.
  • 80K+ users
  • 3,000+ partner organizations
  • 80% less reporting effort
View selected proof

AI-assisted build · prototype

MountFlow

A working vertical SaaS prototype translating a specialist physical workflow into a clear digital operating system.
  • Multi-page SaaS prototype
  • Workflow and domain mapping
  • AI-assisted product development
View selected proof

Evidence before claims

A portfolio should make scrutiny easier.

Every case separates confirmed project evidence, product evaluation, simulated product direction, and target outcomes. Confidential enterprise work is anonymized; all product visuals use public-safe or synthetic data.

Review the full body of work

Built where it matters

AI products that survive contact with the real workflow.

Desheng Lu · Toronto, CanadaEnterprise AI · Product & Portfolio Leadership