Enterprise
AI Hub
Leading discovery, strategy, and UX design for a new Enterprise AI Hub — from stakeholder interviews and personas to a clickable prototype that convinced leadership to greenlight the build. Now in active development.
Role:
Senior Product Designer
Tools:
Figma
Domain:
Enterprise AI · Platform Design
Team:
Inspire11 · US Foods
4wk
Engagement length — from zero to a full UX vision, prototype, and stakeholder-validated platform recommendation
14+
Fragmented AI tools in flight across the org when we started — no governance, no single entry point, no visibility
6
Core platform experiences designed end-to-end in the prototype — Steward, AI Pantry, App Builder, Governance & more
The Problem
AI was everywhere.
And nowhere at once.
US Foods had launched ChefGPT — a custom AI platform — over a year before our engagement. But it hadn't materially evolved since launch. No product owner. No adoption strategy. No governance. And quietly, shadow AI tools were proliferating across teams: Cursor, Synthesia, Copilot, Bedrock, Snowflake Cortex.
The result: employees didn't know what existed, what was approved, or how to get access. And leadership had no visibility into what was running, what it cost, or whether any of it was working.
No product ownership
No discovery Surface
Zero governance
14+ fragmented tools
Shadow AI proliferation
No ROI visibility
1.
ChefGPT stagnated - Built as a proof-of-concept, never transitioned to a product. No roadmap, no owner, no adoption support — and growing user distrust.
2.
Tools multiplied unseen - Sales, HR, Supply Chain, and DigiTech were all running separate AI experiments. No centralized tracking, no approval process.
3.
Governance was an afterthought - No data classification enforcement, no audit trail, no visibility into which models were touching sensitive data.
4.
Leadership couldn't see impact - No usage analytics, no cost tracking, no ROI measurement. AI was a cost center with no defensible value story.
My Approach
Ground every decision in a real human.
Before designing a single screen, I ran stakeholder interviews across Sales, Supply Chain, Compliance, Legal, and AI Enablement. From those sessions, I synthesized four composite personas representing the distinct ways people interact with AI at US Foods — each with different needs, trust levels, and access patterns.
AI Enablement Lead
Jordan · Enterprise AI
Owns the platform strategy. Needs governance tooling, adoption visibility, and a way to track ROI across the portfolio — not just usage counts.
Frontline Operator
Alex · Supply Chain
Task-oriented, time-pressured. Needs AI to help with specific daily workflows — not model selection, prompt engineering, or approval queues.
Compliance & Risk Owner
Morgan · Legal & Compliance
Skeptical of AI without guardrails. Needs audit trails, data classification enforcement, and clear escalation paths when outputs raise flags.
Business Value Owner
Taylor · Sales Leadership
Wants AI to drive measurable outcomes. Needs ROI dashboards, success stories, and tools tied directly to revenue metrics — not experiments.
“Adoption and trust don’t break because people don’t want to use AI. They break when AI creates more work, more risk, or more uncertainty than it removes.”
Journey Mapping
Where AI breaks down
in the real world.
Journey maps matter because they tell us where to intervene — not just what to build. I mapped the end-to-end AI lifecycle across all four personas, from the moment someone hears about a tool to the moment they actually use it. The gaps weren't subtle. Discovery was word-of-mouth. Access required knowing the right person. Governance was an email chain. These journeys became the blueprint for every design decision in the platform.
01 · Discover
Finding a tool
Word of mouth. Slack messages. Stumbling onto SharePoint. No central catalog, no search.
Gap: No discovery surface
02 · Request
Getting access
Email the IT help desk. Wait. Follow up. Unclear who approves what. Often just give up.
Gap: No self-service access
03 · Execute
Doing the work
No workflow templates. No knowledge integration. No guidance on what data is safe.
Gap: No governed execution
04 · Govern
Staying compliant
Compliance teams have no visibility. No audit trail. No way to enforce policies at scale.
Gap: No audit capability
05 · Measure
Proving value
No usage data. No ROI tracking. No way to justify investment or identify what should scale.
Gap: No portfolio view
The Solution
One platform.
Every AI workflow.
I designed a clickable prototype for the Enterprise AI Hub — going well beyond the static wireframes the SOW required. The prototype demonstrates all seven core platform experiences as a connected system: not individual screens, but a coherent product vision. It became the artifact that aligned stakeholders, drove the platform recommendation, and ultimately unlocked the build. The Hub is now in active development.
Steward: AI-powered conversational assistant that serves as the primary entry point for discovering tools, asking questions, and launching workflows across the enterprise.
App Builder: Hybrid wizard that lets power users propose and configure new AI applications through guided conversation and visual form-based setup.
AI Pantry: Searchable catalog of all approved and pilot AI applications, filterable by category, with quick access to details, ratings, and access requests.
Portfolio Dashboard: Executive-level analytics hub showing usage trends, ROI metrics, cost tracking, integration health, and actionable optimization recommendations.
Use Case Intake: Standardized 32-question submission form capturing business goals, AI governance evaluation, and value metrics to route new AI proposals through risk-based review.
Governance: Centralized review console for approvals, audit trails, risk classification, and active guardrail monitoring across all enterprise AI applications.
Platform Design
Six experiences.
One connected system.
Each module was designed to solve a specific gap identified in the journey maps — and to connect seamlessly to the others. The result is a platform where governance doesn't slow things down, it's built into the flow.
01 - Steward
Conversational Entry Point
Task-oriented AI discovery. Users start with what they need to do — not which model to pick. Recommended tools surface based on role and context, with model visibility built in.
03 - AI Pantry
Enterprise Discovery Catalog
A searchable, role-personalized marketplace of every approved AI application. Risk level, usage stats, approval status, and ratings visible at a glance — pilot tools clearly separated from production.
05 - Governance
Compliance in One Place
Pending approvals, active guardrail summaries, and a fully searchable audit trail of every AI interaction — user, prompt, model, data source, and output — all in one workspace.
02 - App Builder
Governed Citizen Development
Power users can create and publish AI applications through a structured, conversational workflow. Guardrails, data access controls, and approval gates are embedded before anything goes live.
04 - Use Case Intake
From Idea to Action
Standardized intake form capturing business goal, data sensitivity, governance requirements, and projected ROI. Low-risk ideas auto-approve. Higher-risk ideas route to the right reviewer automatically.
06 - Portfolio Dashboard
AI as a Managed Asset
Executive view showing adoption trends, cost by department, ROI by application, and integration health. Turns AI from a collection of tools into a defensible, measurable enterprise portfolio.
Outcome
The prototype unlocked the build.
Static wireframes were the requirement. A clickable, fully navigable prototype is what I delivered. Stakeholders could see and react to the platform — not just imagine it. That shift from abstract strategy to tangible experience is what moved the conversation from "what should we build?" to "when do we start?"
The recommendation was accepted. US Foods is now moving forward with the Enterprise AI Hub — and the prototype I designed became the product blueprint the development team is building from.
“Hey man, just wanted to reiterate how impressive the clickable prototype is. That massively exceeded the expectations in the SOW and will certainly show well. Really nice work!”
“Great work man, you’ve really synthesized a lot in just one week and that overview just now was strong.”
“This is a beautiful asset Aaron!”
“Appreciate you going above and beyond the SOW on this, that’s how we move the needle on winning new work.”
“Loved the UI!”
Reflection
What I took away from this engagement.
1.
Go beyond the brief, intentionally
The SOW called for static wireframes. I built a clickable prototype. Not to show off — because it was the right artifact to align stakeholders on a complex, multi-surface platform. The brief is a floor, not a ceiling.
2.
Governance and UX aren't opposites
The biggest design challenge wasn't the interface — it was making governance feel invisible to end users while giving compliance teams exactly what they need. That tension is where the real design thinking lived.
3.
A prototype is the strongest recommendation you can make
No slide deck makes an argument like something you can click through. The prototype didn't just demonstrate the platform — it made the case for building it. When stakeholders can see the experience, the decision to invest becomes obvious.