Blog
Notes on AI that have dirt under their fingernails.
Short, practical essays on model behaviour, reliability, judgement, and the choices that make AI systems useful in the real world.
The Value of AI Expertise: Knowing When to Say No
A practical argument for AI leadership that can reject overbuilt solutions and protect teams from waste.
- AI strategy
- Leadership
- Delivery
From Guessing to Guarantees
A look at abstention-aware evaluation and information budgets as practical controls for hallucination risk.
- LLMs
- Reliability
- Evaluation
Are LLMs Really Deterministic?
A grounded explanation of sampling, implementation details, and why repeat prompts can still drift.
- LLMs
- Systems
- Inference