Thoughts from inside live AI programs.
Notes on governance, delivery, model decisions, workflow design, and what tends to hold up once the work leaves the workshop and enters the organisation.
Claude in a Microsoft-heavy company: the state of play in April 2026
The question isn't whether AI can help inside a Microsoft-heavy environment. It is whether you are willing to do the work required to make it reliable.
OpenClaw and the next agent interface: execution, risk, guardrails
Some projects change how you think about agents in five minutes. OpenClaw is one of them.
AI transformation for regulated firms: hype vs reality
The shift from AI-sceptic to AI-believer doesn't happen at a demo. It happens when a real workflow runs end-to-end and you can answer the questions that actually matter.
Claude Opus 4.6: what regulated firms need to change now
When the model starts behaving like a junior operator, the question isn't "should we try AI?" It is: are we ready for AI that actually does work?
Build your AI muscle: 6,000+ hours of practical lessons
The shift from "can it write an email?" to "can I use this under pressure?" is the whole game.
Culture eats AI strategy: regulated deployment lessons
The best AI strategy is a set of sensible guardrails, then an overinvestment in AI culture.
From shadow AI to approved patterns: a principal playbook
Shadow AI isn't a future risk. It is already in the building. The question is what you put in its place.
Micro-evals for fast model change: a licensee playbook
Vibe-based AI governance falls apart when models update monthly. Micro-evals are the cheap way to stay oriented.
Picking the first AI workflow: the three-day rule
If you can't ship the first workflow in three days, you picked the wrong first workflow.
AI readiness for regulated firms: what changes now
If "this will reshape the industry" still sounds like tech theatre, the reference point is the problem.
AI productivity without burnout: the AI vampire problem
There is a moment with agentic AI that feels incredible, and a second moment nobody mentions in vendor demos.
The Audit Trail Test: supervision-ready AI for compliance teams
A thought experiment every compliance and operations lead should run before any AI workflow goes live.