Utah-admitted attorney, nearly four decades inside legal knowledge and technology. Today I design and build governed, agentic-AI operating systems for legal practices — from the solo attorney to the global firm.
Start a conversation See the 50-state fanoutI sit where attorneys and engineers meet — and I'm trusted by both. I turn generative AI from a demo into day-to-day practice: repeatable workflows, precedent and playbook libraries, and the governance that keeps the output verifiable.
Agentic-AI workflows and a transferable operating model — multi-agent orchestration that produces attorney-ready work, not novelties.
Citation-checked, source-verified, compliant by design. The governance and knowledge architecture are built in, not bolted on.
From the solo attorney to the practice group to the global firm — the same governed engine, fanned out and adopted.
A working demonstration of governed legal AI at scale: a single, locked golden template, fanned across all 50 states and the District of Columbia by an AI swarm I orchestrate in Claude Cowork. One voice, fifty-one jurisdictions, every claim held to its primary source — re-run fresh for each issue.
Click any cover to open the full, source-verified brief — one golden template, fanned to all 51.
A single golden example sets the voice, structure, and quality bar before anything fans out.
One natural-language orchestrator — Role, Inputs, Work loop, Rules — runs the swarm. No brittle scripts.
The swarm fans the stencil across jurisdictions in parallel; the folder is the kanban — state lives in file existence, not a database.
Every statute, bulletin, and figure is checked against primary law; a representative slice is sampled to keep quality defensible at scale.
I've built and led knowledge-management and practice-automation functions inside global firms — creating programs from scratch, integrating systems with real attorney workflows, and earning adoption by tying every initiative to measurable value. At Hunton I built 4 practice templates into the Client Workroom that scaled adoption to 8,000 users across more than 500 client entities.
Tool-agnostic field notes from AI-enablement work — the “fights with the AI” and what building and shipping governed legal AI is actually teaching me. Not legal advice.
Eighty-four case citations checked for later treatment in one evening — and the three habits that kept the speed safe: ask for the batch, confirm the case, read the opinion behind the flag.
The AI handed me 356 quotations from official sources and twenty-eight were not on the page — the boring step that caught them, the seven court decisions a summary got wrong, and the words a link check waves through.
I wrote down the assumption that could sink the project — then graded a different one, because it was the half that could be measured. What saved me was an AI that refused to be helpful.
If you want it done with both legal credibility and hands-on AI capability, let's talk. Based in Richmond, Virginia; set up for Washington, DC on a hybrid basis.