LexAssist: Cutting Through Enterprise Ops Chaos – One Request at a Time
In high-pressure ops environments, clarity isn’t nice-to-have. It’s non-negotiable.
LexAssist was built to cut through fragmented workflows, not just automate them. It was about designing a system that could hold context across functions—contracts, tickets, finance threads—and make sense of them without needing users to spell everything out.
Context First
Project Overview
LexAssist is an AI-powered support agent purpose-built for complex enterprise operations.
The product vision was to create a system that could handle multi-threaded enterprise requests with speed, nuance, and context-awareness. This meant extracting relevant information from contracts, surfacing key clauses, tying conversations to approval workflows, and ensuring continuity across threads—without the user ever needing to repeat themselves.
The platform combined conversational AI, workflow orchestration, and enterprise-grade data visibility to deliver trust, not just responses.
"A Basic question leading to five open tabs, three handoffs, and zero answers? That's a big NO!"
My Role: Lead Product Designer/Product Manager
As both Lead and Individual Contributor on this initiative, I was responsible for driving the design vision and execution across the product lifecycle:
- Strategic Product Definition — Collaborated with Product and Engineering to define the platform’s value proposition, product pillars, and AI differentiators
- User Research — Led contextual inquiries, interviews, and co-creation workshops with 50+ users across Finance, Ops, Legal, and IT
- Persona & Journey Mapping — Created archetypes ranging from frontline agents to compliance leads, mapping their needs across typical workflows
- Information Architecture — Defined a modular system that supported horizontal escalation, state retention, and cross-departmental workflows
- Interaction Design — Designed multi-layered flows that adapted to task urgency, required approvals, and conversation context
- UI & Design System — Developed an enterprise-ready design system emphasizing legibility, authority, and flexibility
- Prototyping & Iteration — Built high-fidelity prototypes in Figma for usability testing and stakeholder demos
- Usability Testing — Ran three rounds of testing to identify sticking points and refine AI affordances
- Cross-Functional Collaboration — Embedded with devs, AI leads, and legal SMEs to ensure accurate implementation of complex business logic
- Design Evangelism — Regularly presented progress, rationale, and tradeoffs to senior stakeholders and exec sponsors
Project Impact
Even before full deployment, early metrics from pilot teams were promising:
- 35% Reduction in time-to-resolution for high-friction workflows
- 80% Adoption rate within first 3 months in Ops & Finance teams
- 25% Boost in collaboration efficiency across teams
- Fewer Errors Manual copy/paste and approval misses were significantly reduced
- Increased AI Engagement 60% of users interacted with AI suggestions regularly
Framing the Challenge
Enterprise users don’t ask isolated questions—they reference invoices, cite contract clauses, and follow threads buried in SharePoint. They're not trying to “chat.” They're trying to get something resolved—fast, and under pressure.
Legacy tools weren’t built for this. They either flood users with rigid flows or bury intent under backend jargon. What you get is:
- Cognitive friction from rigid flows
- Ping-pong escalations across departments
- Lost trust every time the system misses the point
Discovery & Research
We kicked off with deep discovery across four enterprise departments—Finance, IT, Legal, and Operations:
- Stakeholder Interviews: Multiple sessions with department heads and IT leaders to uncover strategic gaps
- User Shadowing & Interviews: 20+ users observed and interviewed to map live workflows and find friction points
- Competitive Audit: studied several key enterprise support tools (from Intercom to ServiceNow) to benchmark strengths and blindspots
Key Insight
Teams weren’t asking for a chatbot. They needed an intelligent teammate—one that could::
- Hold Context across approvals and documentation
- Resolve, not escalate most inbound queries
- Stay invisible when possible and accountable when needed
We needed operational simplicity. Pronto!
1
Track user intent across threads2
Handle layered, messy queries3
Respond with clarity, not canned replies
Empathy @ Scale
Defining the Opportunity
Beneath every support ticket was a blocked intention:
- A contract stalled waiting on sign-off and approvals
- A task flow put on hold due to missing file context
- A request delayed by fragmented systems and reroutes
What we heard wasn’t “make this easier.”
It was: “Why is something simple taking five tools and ten steps?” These users didn’t need another interface, but needed a peer – one that could cut through the noise before it turned into delay.
- Item 1
- Item 2
- Item 3
"What are users really trying to do—and why can’t they get there today?"
Opportunity Framing
- Unblock intent: Surface what users are trying to achieve—not just what they ask.
- Design for momentum: Build a UX layer that accelerates workflows, not just reacts to them. In other words, design for velocity, not just accuracy.
- Introduce the Workflow Agent: Not a chatbot. A context-aware copilot, purpose-built for enterprise messiness.
- Item 1
- Item 2
- Item 3
1
Reads context, not just commands2
Surfaces next steps before you're stuck3
Handles nuance without the hand-holding
Workflow Thinking
Strategic Design Objectives
In high-stakes workflows, aesthetics don’t matter if the system stalls, confuses, or hides the next step. Every interface choice here served one purpose: make decisions easier to take, faster to trace, and harder to fumble.
Design Priorities
- Minimize hesitation. Users should never second-guess where to click or what happens next. Make it obvious.
- Anticipate intent. The agent needed to recognize patterns and act before the user had to ask.
- Audit without friction. Every interaction left behind breadcrumbs—clear, reviewable, and reversible.
This Design endeavor was about making decisions easier, faster, and more defensible across the organization.
- Item 1
- Item 2
- Item 3
Interface in Service
Interaction System + Interface Design
Form matched urgency. State matched need.
This wasn’t a static dashboard – it flexed based on operational pressure. Escalation paths, low-friction exits, and multi-stage flows weren’t designed for elegance. They were designed for survivability in enterprise ops.
We designed a system that responded to what the user needed:
- A critical escalation surfaced a linear, focused flow—with confirmation steps and risk indicators.
- Routine queries stayed lightweight—contextual buttons, smart defaults, and frictionless exit paths.
- Multi-step tasks stacked horizontally with persistent anchors—so switching modes didn’t mean starting over.
Instead of styling for clarity, we structured for it. Cards, threads, and layout patterns weren’t decorative—they were functional markers. They told users: here’s where you are, here’s what’s at stake, here’s what comes next.
- Item 1
- Item 2
- Item 3
1
Clear next steps – no guesswork2
Smart defaults reduce clicks3
Layouts guide, not distract
Learnings
Outcomes + Takeaways
