Portfolio · AI
AI that frontline teams can actually use.
I design AI-powered products for enterprise teams: care agents on a live call, network engineers, people drafting from dense documents. Getting the model to answer is usually the easy part. Making the answer useful to the person reading it is the job.
These case studies cover the research, the rules and the evaluation that got each one there.
T-Mobile · Research case study
NORA
Turning root-cause analysis into answers a care agent can use.
Frontline care agents were getting accurate diagnostics written for engineers, not for a customer on the line. I ran interviews and observation, built a persona and a Playbook of rules, then validated and evaluated the AI output that followed them.
15
interviews & observations
5
design principles → a full Playbook
T-Mobile · AI design system
NetHive IQ
One AI, three very different jobs.
An AI layer was moving into every corner of T-Mobile’s network operations platform, with no shared standard for how it should sound or when it should act on its own. I built the Voice & Tone Style Guide, AI Best Practices, and Explainable AI standards that tie it together.
3
user groups, one AI voice
3
living guides: voice, practices, XAI
AT&T · Usability study
Ask AT&T Docs AI
Making Compare and Draft Creation workflows click.
Before rolling out two new AI capabilities — comparing documents and drafting from that comparison — I ran a moderated usability study with 13 participants to see whether people could actually find and use them.
13
participants, 11 tasks
3
tactical recommendations shipped to design
Building an AI tool people have to trust?
I can walk through how any of these were researched, written and tested, including the parts that didn't make the write-up.
Get in touch