Thomson Reuters

Earning legal experts' trust in agentic AI through validation experiences grounded in human judgment.
Earning legal experts' trust in agentic AI through validation experiences grounded in human judgment.
At a glance
At a glance

ROLE

Lead UX Designer, AI Experience & Enablement · Thomson Reuters (in-house)

TIMELINE

1 year · 2025-2026

Team

Led a senior designer, partnered with a principal designer, and managed stakeholder relationships end to end, from strategy through execution.

SCOPE

Agentic AI content pipeline · Human-in-the-loop validation experience · Legal editorial workflows

I OWNED

End-to-end UX and visual design for the validation experience, from strategy and cross-functional alignment through frameworks, research collaboration, and shipped design.

The challenge

Redesign how legal content gets produced when agentic AI does the heavy lifting, in a domain where an unvalidated error carries real consequences.

What I did

Led UX for human-in-the-loop validation experiences, authored the Human-AI Interaction Design Checklist for responsible AI practice, and built a reusable AI skills ecosystem with a cross-functional contribution model.

The outcome

Validation experiences legal experts rated highly for ease of use and trust, and a design framework for responsible AI practice.

TL;DR

When AI agents start producing work that lawyers depend on, the design problem stops being the interface and becomes the judgment: where humans validate, how they stay in control, and what earns their trust. As Lead UX Designer on Thomson Reuters' AI Experience and Enablement team, I led design for the human-in-the-loop validation experiences at the centre of Thomson Reuters' agentic AI transformation of legal content workflows.
Continued on request
The full case study is available on request.
If we've talked shop, you have the password. If not, that's easily fixed.
Request access by email

My takeaway — and what that means for you

In AI products, trust, transparency, and collaboration aren't features of the design — they are the design. If your experts are quietly re-verifying everything your product does, that's the validation experience problem worth solving first.