Decision models: AI built to decide, not to write
Models built to make a small judgement rather than write text, with an LLM's flexibility and a classifier's speed and cost.
Read articleBlog
Practical writing on AI, machine learning, NLP, and data science. What works, what doesn't, and how to think about applying it to real business problems.
RSS feed
Is AI good enough yet?
You want the benefits of AI without the pain of a failed project. But how do you tell which jobs it's ready for, and which need more investigation?
Read article
Agents Aren't Chatbots With Extra Steps
Plenty of AI tools are being called "agents", but what can you actually delegate to them? And when would a simpler workflow do the job just as well?
Read article
The question that predicts what AI does to your job
AI exposure rankings miss the real split: generation vs verification. Who still owns the decision predicts what happens to pay and hiring.
Read article
Same process, new motor: what history says about the AI productivity paradox
If AI is everywhere, why isn't it in the productivity numbers? The electrification of factories holds the answer: substitution is not reorganisation, and most organisations are still running group drive.
Read article
Why Passing Standard AI Safety Benchmarks Is Not Enough for Enterprise AI Systems
You can pass every AI safety benchmark and still break your own rules. Org-policy testing is the missing layer most teams skip.
Read article
Why Enterprise AI Looks Like Transformation But Isn't
Only around 15% of enterprise AI use cases generate measurable ROI. The reasons are almost always organisational rather than technical. This piece looks at why, how to spot the warning signs, and how to think about doing better.
Read article
AI and Expert-Level Work
OpenAI's GDPval benchmark shows that top AI models now match expert-level quality on nearly half of complex professional tasks. Here's what the data actually shows and what it means.
Read article
The Same Mistake, Different Technology: What MIT's AI Study Really Shows
MIT's study found 95% of enterprise AI pilots fail to deliver measurable ROI. After 20+ years in AI, Paul Ferguson explains why this isn't surprising — and what organisations can do about it.
Read article
RAG vs Fine-Tuning for Enterprise LLMs: A Practical Guide to Choosing the Right Approach
A step-by-step framework for technical leaders choosing between RAG and fine-tuning for enterprise LLMs, including when to combine both in a hybrid approach.
Read articleLet's Talk
Have a question about AI?
Book a free 30-minute call and we will answer it honestly.