Case Study

Agiloft

About/ Agiloft is a KKR-backed contract lifecycle management platform used by large legal and procurement teams.

Agiloft acquired a promising AI product and made a strategic decision to build its new AI platform around it. The challenge: re-design the product from the ground up and get it launch-ready in six months.

Role/ VP, Product & Experience Design
Timeline/ Aug 2025–Present
Category/ Enterprise AI · Product Design
Mandate/ Lead the design of Astra, Agiloft's new AI platform.
Team/ 3 Designers • 3 PMs • 9 Engineers

6 Months

Launch-ready enterprise platform: full component library and all UX flows.

2 Assistants

AI-powered legal assistants designed end-to-end.

WCAG

Full accessibility compliance built into the design system.

1 System

Unified design system powering all product surfaces.

The entry point for an AI assistant. Designed to feel approachable without sacrificing the gravitas enterprise legal workflows require.

The entry point for the Astra AI assistant. Designed to feel approachable without sacrificing the gravitas enterprise legal workflows require.

The product's home screen. The IA surfaces Agents and Workflows as first-class navigation — making AI capabilities feel native, not bolted on.

Astra home screen. The IA surfaces Agents and Workflows as first-class navigation, making AI capabilities feel native, not bolted on.

An AI agent mid-task. Structured responses, contextual action buttons, and a conversational model that guides users through complex negotiation workflows without overwhelming them.

An AI agent mid-task. Structured responses, contextual action buttons, and a conversational model that guides users through complex negotiation workflows without overwhelming them.

High-density enterprise data made accessible. Risk level badges, compliance scores, and pass/fail indicators let legal teams scan across dozens of documents at a glance.

High-density enterprise data made accessible. Risk level badges, compliance scores, and pass/fail indicators let legal teams scan across dozens of documents at a glance.

01

Design system first

With only six months to design an entire platform, consistency couldn't be something we cleaned up later. Establishing tokens, components, accessibility, and interaction patterns first let the product scale quickly without fragmenting as three squads built in parallel.

 

02

Parallel with engineering

Ran design and engineering in tight parallel to hit a six-month window without sacrificing quality or compliance. We created three squads to divide the work thematically and encourage close collaboration. This kept handoff friction low and reduced rework as the product took shape.

 

03

AI as a first-class product decision

We defined the human–AI interaction model early, including how agents communicate progress, request decisions, surface risk, and hand control back to the user. That model became foundational to Astra's high-stakes legal workflows.