Insights
Research that works on Monday morning.
Practitioner research on context engineering, AI governance, builder bias, and how people actually use technology.
AI Strategy
For executives navigating the AI mandate. Start here if you need a plan.
The First Thing We Protect Is Your People
2026
Psychological safety is not a soft skill. It is a strategic requirement. The strategy does not work if the people executing it are afraid.
How to Evaluate an AI Strategy Firm (Including Us)
2026
Seven questions to ask any AI strategy firm before signing. We publish this because the best way to earn trust is to help you evaluate everyone honestly.
What AI-Ready Actually Means
2026
Three things AI-ready actually means. None of them require a platform purchase. Cut through the vendor noise.
Your AI Pilot Failed
2026
Three failure modes behind every failed AI pilot. None of them are technology problems. All of them are strategy problems.
The 90-Day Window
2026
Three clocks are ticking on your AI strategy. Confidence, competition, and budget. Every month without a plan costs more than you think.
Start Here
The most accessible entry point for practitioners.
Strategy
The big picture. What kind of change are we talking about, and whose assumptions are driving it?
Remodel, Gut Rehab, or New Construction
2026
Three types of AI transformation. Construction figured this out centuries ago. Three risk profiles. Three honest conversations about what stays and what goes.
Context Engineering: The Builder's Bias Problem
2026
Every technology builder carries a personal stack that creates invisible assumptions. Context engineering is the discipline of governing whose perspective shapes intelligent systems.
Design
How should AI actually help people? What should the interface feel like?
We Organize AI Around People, Not Humanize AI for People
2026
Structural organization compounds. What it looks like to build AI around human intent — and why that produces fundamentally different architecture.
Interface Endpoints: The Public API Is a Feeling
2026
The interface between a complex system and a human should be a feeling, not a dashboard. Three dots. Glance, drill down, done.
Architecture
How the system learns, remembers, and gets better over time.
Crystallization
2026
Most systems treat unresolved questions as tasks to assign. What if the question did the work? Standing queries with geometric shape, filtering the stream until the answer precipitates on its own.
Building an AI That Digests Its Own Experience
2026
Most AI agents act, return results, and forget. We built one that prunes what didn't work, archives what did, and gets faster — not just busier.
Governance
Staying in control. Protecting data. Making sure "stop" means stop.
Simple Override Beats Elegant Architecture
2026
The case for hard human control in local AI communities. Why a single human override is more valuable than any sophisticated governance circuit.
The Access Layer Pattern
2026
How to use AI on real business data without sending secrets to the cloud or breaking the model's reasoning.
Deep Reads
The full design philosophy and technical foundations.
The Core Instinct
2026
Open loops conserve momentum. The access layer resolves entropy. Experiences real-enough ARE real. Design convictions for building AI systems that metabolize, not file.
When Recursion Gets Cheap
2026
The organizational primitive that only worked for militaries and corporations now works for one person on a laptop. What changes when forking, spawning, and coordinating cost nothing.
Research Prompt: Context Engineering
2026
The prompt used to generate the Context Engineering paper. Published so others can generate their own version.
Prefer it packaged for your team?