Walmart

Squiggly - Walmart’s unified Associate AI platform

David helped shift associate AI from a growing set of disconnected agents into Squiggly - a governed platform of shared identity, interaction patterns, knowledge architecture, reusable components, and trusted experiences designed for 2.1M associates worldwide.

The shipped Squiggly platform - one coherent associate AI experience.
Role
Group Director, Associate Experiences
Timeframe
2024 - 2026
Scale
Designed for 2.1M associates worldwide
Team
Cross-functional product, design, engineering, and enterprise leadership

Executive read

Treating associate AI as governed infrastructure - not another assistant - is what made it trustworthy and reusable at Fortune-1 scale.

AI capability was spreading faster than enterprise coherence. Without a governed platform, each new agent risked adding another name, behavior, sign-in, trust model, and support burden.

The reframe was to treat associate AI as infrastructure, not another assistant - make trust a product requirement, define reusable patterns across surfaces, and set the operating-model direction, so capability could be built once and reused coherently across field and home-office.

What follows is the shipped system and its staged reach across 2.1M associates - reach and rollout, not adoption claims.

Reach and rollout

  • 2.1M associates worldwide - the platform design target
  • 120K Sam’s Club US field associates - v1 released
  • 1.6M Walmart US field associates - v2 rolling out
  • v3 home-office associates - underway

Reach and rollout figures only - not adoption, usage, ROI, or business impact.

Tension

Fragmentation had become the experience

Associates were not meeting one intelligent enterprise. They were meeting many disconnected tools - each agent with its own name, behavior, and trust signals. As more AI arrived, the experience risked becoming less clear, not more capable.

The reframe started here: the work was not to design another assistant. It was to make AI understandable, trustworthy, and reusable across the enterprise.

Decision

Trust before novelty

The strategy led with trust, not novelty. The signature stance was simple: AI should augment human decisions, never replace them - and that principle shaped the product, the interaction model, and the brand.

Rather than a scatter of separate tools, the work moved toward a single, recognizable platform that could earn confidence over time.

AI should augment human decisions, never replace them.

Reframe

From touchpoint to a governed platform

My Assistant began as one of many AI touchpoints. The strategic move was to stop treating it as a destination and start treating it as part of a governed platform - a system of reusable building blocks that new use cases could plug into, instead of each team reinventing AI.

One system served different contexts: home-office associates through My Assistant, field associates through Sidekick. This moved the work from interface design to enterprise product architecture.

One system beneath many surfaces - reusable building blocks across My Assistant (home office) and Sidekick (field).

System

Confidence had to be visible in the interaction model

Trust could not be a brand claim. It had to appear in the interaction itself - reasoning visibility, source citations, verified-data labeling, and a verify-before-submit step on higher-stakes actions - so associates could see where an answer came from and when a human checkpoint was required.

For the frontline, that confidence also had to survive the physical reality of the work: shared devices, limited attention, and noise. The experience moved from “chat with a tool” toward “get to the outcome safely and quickly.”

Trust as interaction: reasoning visibility, citations, verified-data labeling, and verify-before-submit.

Evidence

Built for Fortune 1 reach

Squiggly v1 was released to all 120K Sam’s Club US field associates. v2 is rolling out to 1.6M Walmart US field associates. v3 is underway for home-office associates. The platform is designed for Walmart’s 2.1M associates worldwide.

These are reach and rollout figures - the audience the platform serves and the status of release - not adoption, usage, or ROI.

  1. v1 Sam’s Club US field · 120K associates Released
  2. v2 Walmart US field · 1.6M associates Rolling out
  3. v3 Home-office associates Underway
  4. Scope 2.1M associates worldwide Design target
Staged rollout by audience. Reach and release status only - not adoption.

Lesson

The platform is trust, not the chat box

When AI capability is no longer the constraint, the real work becomes trust and coherence - the difference between launching another assistant and building an enterprise capability.

The next frontier is governing this as an operating model, so many teams can contribute to one coherent system. That direction is set; building it out is the work still ahead.

Work together

Have a problem worth getting right?

For executive leadership, advisory, speaking, or collaboration around enterprise design, product strategy, intelligent systems, people experience, or complex organizational change.