Long-form writing on AI, government, FCA practice, procurement, and what we learn watching backlog graphs go down. Written by people doing the work, not a content team.
Frontier model demos make great press. Public-sector decisions need predictable, observable, replayable systems. Here's what we mean by "boring," why it costs more to build, and why agencies should refuse to settle for less.
By Maya Okonkwo · 14 min · April 2026
Read the essay →The arithmetic of state and federal hiring vs. citizen demand.
M. Okonkwo · Apr 2026
What the disability backlog clearance actually felt like, day by day.
P. Nair · Mar 2026
What changes for relator firms, for the DOJ, and for defendants.
D. Reyes · Mar 2026
Why we made every model call deterministically replayable, and what it cost.
P. Nair · Feb 2026
Six clauses every state RFP gets wrong.
M. Okonkwo · Feb 2026
The audit trail bar and how to clear it.
D. Reyes · Jan 2026
End-to-end series on disability initial determinations, from form to decision letter.
The org-chart, control-by-control reality of authorization.
How we made agents observable, replayable, and diffable.
How AI document review reshapes economics for plaintiff firms.
CEO. Writes about the agency-side of GovTech.
President & GC. Writes about FCA, defensibility, ethics.
CTO. Writes about engineering, document AI, agents.
Head of Federal. Writes about procurement & capture.
Forward Deployed. Field notes from agencies.
State CIOs, IGs, and counsel. Edited but not sanitized.
The post that named the framework most agencies now use.
An eight-step playbook with the metrics that matter.
And what to demand instead.
If a post is for the algorithm and not the reader, we kill it.
We compare on substance. Never with snide.
No customer is named without explicit permission.