
Designing an AI-Powered Clause Library
Zero-to-one AI-driven feature set for a legal technology SaaS startup.

Bringing user insights into product strategy since the '00s.
Organizations that move fast and pivot successfully aren't guessing. They have someone turning user evidence into insights that enable decision-making in uncertain conditions. I'm that person.
Most research functions produce reports. I build organizational capabilities to deliver the insights that drive tactical and strategic decisions.
The distinction matters. A report is a point-in-time artifact. A capability is a set of processes, relationships, and communication channels that run continuously and bring user evidence into the rooms where the consequential decisions are made.
That means I usually work at three levels in parallel:
I also bring something most UX researchers don't: over 25 years of experience researching how people work with computers and automation. My PhD work investigated how pilots and air traffic controllers calibrate trust in and build mental models of automated systems.
That same problem is now the central design challenge of every AI-augmented product. I've researched it from both sides: designing AI features that users can understand and trust, and integrating AI into research operations so humans stay in the interpretive loop. This isn't a recent interest. It's the thread that connects everything I've done.
Case studies from recent product research and strategy work.

Zero-to-one AI-driven feature set for a legal technology SaaS startup.
Mixed-methods product-market fit research.
Accessibility evaluation and program-building.
Rapid discovery and design for a mobile-first predictive analytics application at one of the world's largest copper mines.
Generative field ethnography to identify personas and unmet needs for a next-generation navigation experience.
My methods span the full research spectrum: discovery and design sprints, contextual inquiry, in-depth interviews, jobs-to-be-done analysis, journey mapping and service blueprinting, persona definition, usability testing, survey research, behavioral analytics, and more. I pick the method that fits the question, not the other way around.I pick the method that fits the question, not the other way around.
I've worked across healthcare, fintech, legal tech, insurance, telecommunications, procurement, and enterprise SaaS. The domain matters less than the complexity. I do my best work in products where the problem space is tangled, the stakeholders are many, and the path from evidence to action requires translation across teams.
I used AI to help me create the content on this site. Did I write every word? No. Did I review, brainstorm, and iterate on the content with some AIs? Of course. Why wouldn't I? Web copy has a distinct rhythm and tone. It's punchy and effortlessly authoritative. That's not how I write. Or think. So of course I used AI for some of the site copy.
Our oldest kid tells me that I should've written all this myself without relying on the clankers. I get it. But I also like to work fast and ship. So I jammed on the content with Opus 4.6, had it build an easily-maintainable site, and stood up v1 within about 24 hours from the first prompt. It's a tool. One I found helpful.
Again, I get it. I understand that current AI offerings are just uncomprehending stochastic parrots that we grant a sometimes-concerning amount of agency. And data centers are the next socio-ecological disaster-in-progress. I feel like I'm making typical Gen-X excuses here, and maybe I am. But what am I gonna do? Not use AI? In this economy? You first. ;-)
I wrote that disclosure. And this sentence. And now I'm trying to avoid an infinite loop.
I'm currently exploring principal, staff, and director-level research roles at companies building complex products. Open to longer-term strategic consulting roles as well. Reach out if your team is working on wicked problems where rapid and valid research needs to be wired into how you make decisions.