I work on AI governance and compliance.
Through Ex Machina Solutions I help organizations decide what to deploy, build the governance program around it, and evaluate whether the system works: adoption roadmaps with named owners, risk and compliance frameworks aligned to NIST AI RMF and ISO/IEC 42001, and evaluation methods with measurable outcomes.
Before this I spent two decades running regulated, compliance-critical programs, including a $1B+ program operating across 64 jurisdictions.
Work
Software and tools I build and operate.
Gavelin
A legislative-intelligence platform I founded and operate, with bill coverage across all 50 states and DC, and speaker-attributed hearing transcripts in nine states at 98.8% attribution accuracy across 1.4 million segments.
AI Website Audit
Enter a URL and get a graded read of how an AI system interprets the site: readability, data accessibility, structure, and content clarity.
The Gilded Age of AI
Roughly 110 milestones of AI moving into public life, set against what each era predicted next. The claim: this is an abbreviated gilded age, and we are nearer its tipping point than the discourse admits.
Background
I spent two decades running regulated, compliance-critical programs. As a state compliance officer I owned quality control, corrective action, and federal-regulator relations for a $1B+ program operating across 64 jurisdictions, cutting open corrective action plans from roughly fifteen to under five and holding the program's payment error rate below 6%. That program was SNAP, administered for the state of Colorado. Before and after it: county-level outreach management, executive direction of a $5–9M organization that raised more than $30M across philanthropy and government, and legislative work that helped pass 25 pieces of state legislation.
Ex Machina Solutions LLC is the advisory practice I run today, for AI governance, risk, and compliance: exmachinasolutions.ai.
Writing
Essays on institutional design, policy operations, and what happens when a system meets the people it was built for. New work lands at The Daimon first; this site is home base for the pieces and their interactive companions.
Why Would Bureaucracy Matter More than Hunger?
The question underneath SNAP's payment error rate, with an interactive state-by-state breakdown behind it.
The SNAP Trolley Problem
A six percent error threshold, a forty-three-day shutdown, and the impossible choice built into an anti-poverty program nobody designed.
Ninety-Eight Percent of Americans
The gap between achievable police reforms and the slogan the movement chose, resurfacing in Wisconsin.
The Case for Nonprofits to Quickly Engage with AI
An argument for mission-driven organizations to stop hesitating and take part in shaping how AI gets built and deployed.
Research & data journalism
Earlier interactive work on federal benefit programs. Every piece stays live at its original URL.
SNAP Payment Error Rates, FY2025
Companion to the bureaucracy-and-hunger essays. The claim: the payment error rate measures administrative capacity and program access, not fraud, and the new penalties pit state budgets against hunger itself.
SNAP Restriction Waivers
A quiet federalism fight over what food assistance can buy, traced through USDA's own approval letters and the court ruling that halted five states mid-stride.
Community Food Insecurity Report
A full diagnostic for any U.S. county, city, or town. The claim: rigorous community research can be automated honestly, if findings that fail adversarial verification are dropped rather than softened.
