Intel
Writing
Field notes on AI, federal technology and building products. Start with the newest below if you are new here.
More: how I work.
New here? Start with four places.
Start here
A Software Factory Finishes What an AI Builder Starts
An app builder gives you a working demo in an afternoon. The last stretch needs a spec, small tickets and gates, and that loop is the product.
AI
46 essays
Defense
29 essays
Tech
22 essays
108 articles · page 1 of 5
Eighteen Months, One Phase One, No Product
What eighteen months of building for verifiability actually bought: four filters for noise, a feasibility study, and no product to show.
The Warfighter Order Is a Lever, Not a Ban
Executive Order 14372 does not ban buybacks outright. It makes capital allocation a past-performance issue the Secretary of War can use.
2025: The Year AI Became Infrastructure, Not Magic
The demos stopped being the deliverable. Cost, evaluation, and integration became the work, and that is what made AI usable.
The Quantum Deadline Is Real. It Is Not 2026
NSA's schedule puts new national security systems on quantum-resistant algorithms by 2030, not next year. The hard part is cryptographic agility.
The Patent Holiday Is the Easy Part
DoD is offering no-fee licenses to a batch of its patents. The royalty was never the reason these technologies sit on a shelf.
Synthetic Data Does Not Solve the Classified Problem
Generating the data is the easy half. Proving it did not inherit a classification and that a model learned from it is where programs stall.
What Allianz's Claims Agent Actually Automates
Allianz now runs an agentic AI on small damage claims. The lesson for federal claims shops is the escalation boundary, not the model.
Price per Token Is the Wrong Number for Model Selection
Anthropic cut Opus pricing by about two thirds. The cost that decides a deployment is still the completed task, not the token.
Agent 365 Is an Inventory Problem Wearing a Dashboard
Agent 365 gives every agent an identity and a policy. That only works if the registry is complete, and agent inventories decay fast.
CMMC Phase 1: What the Rule Actually Requires
The CMMC clause took effect on November 10, 2025, but DoD kept three years of discretion. The date matters less than the first bid you cannot submit.
Section 889 Is an Evidence Problem, Not a Tool Problem
The prohibition covers five vendors and two dates. No visibility platform can prove a bill of materials its suppliers will not disclose.
Agentforce Works Because the CRM Was Already Governed
Salesforce's agents inherit an object model and a permission model that predate them. That is why they work, and why they stop at the CRM.
AI Sustainment Is Performance Sustainment
The RDT&E to O&M shift exposes a gap: O&M contracts are scoped for stable systems, and an AI system is never stable enough to leave alone.
Human-Machine Formations: The Constraint Is the Operator
The Army has the robots. What it does not have is a plan for the people who supervise them, and that is the program's real bottleneck.
End of page 1 of 5. The archive continues.