AI Engineering
Why most enterprise AI pilots never reach production
It is almost never the model. Five infrastructure blockers, and how to clear each one.
Coming soon
Insights
Engineering perspectives, not thought leadership.
Specific enough to be useful to someone doing the same work, and honest about what did not go to plan.
AI Engineering
Building RAG systems that survive an audit
Data lineage, permissions enforced at retrieval time, and the evaluation harness you need before launch.
Coming soon
AI Engineering
Agentic workflows: where the permission boundary belongs
Tool-calling agents are only as safe as the credentials you hand them. A practical design pattern.
Coming soon
AI Engineering
What an AI evaluation harness actually contains
Golden datasets, regression suites and scoring that catches a quality drop before your users do.
Coming soon
AI Engineering
Cutting LLM spend without cutting quality
Routing, caching, context budgeting and model selection — where the real savings are.
Coming soon
Cloud
Landing zones that survive contact with an auditor
Account structure, guardrails as code, and the logging design you'll wish you had chosen first.
Coming soon
Platform
You probably do not need a service mesh
A decision framework for when the operational cost is justified — and the cases where it isn't.
Coming soon
Security
The Active Directory findings we see in almost every test
Kerberoasting, delegation and certificate services. What they look like and how to close them.
Coming soon
Want one of these sooner?
Tell us which topic would be useful to you and we'll prioritise writing it.