From the field
Practitioner-written guides on rates, scoping, delivery, and client management. Built from real engagements, not theory.
Consultant, BA, PM, developer, solutions engineer — one seat, and a reported pay range from £36k to over $1m. What the role actually is, who's hiring, and the test for whether you were promoted or just consolidated.
OpenAI, then Anthropic, now Meta. Four disclosures in sixteen days — two of them the same vendor misconfiguration. What the incident reports say once you read the technique lists.
What is a certification for when you can open the model in another tab and have it answer questions about itself? An honest account from someone who holds several.
Defending against agents with agents is a paradox: the defender is the same animal as the threat, differing only by instruction. An honest take, with our own probe data.
The model serving a third of production traffic has a shutdown date. Popularity is measured everywhere and mortality nowhere, so we joined the two.
The workbench shifted: deliverables now get built inside agentic tools. Here's what's really in the bag in 2026, with the reasoning, and an honest admission the survey will correct.
Most of our readers are in the US and most rate guides are written for somewhere else. Here's what American engagements are actually clearing in 2026.
Most of this site's traffic isn't human. Here's what our own Bot Ledger taught us about how consultants get found in 2026, and what to do about it.
AI models are being retired faster than most contracts run. Here's the clause that stops a vendor's sunset schedule becoming your unpaid migration project.
Over a hundred models repriced in a single month this year. Here's how to quote AI work without betting your margin on a vendor's pricing page.
Data problems kill more AI projects than any technical failure. Here's the two week audit that finds the bodies before you've signed up to deliver on top of them.
The high risk obligations are landing now, US clients are discovering the Act applies to them, and somebody in every engagement has to be the adult. Here's the working consultant's version.
Everyone's got an opinion on what AI consultants should charge. Here's what the market's actually paying.
The official reason is always 'scope creep' or 'change in priorities'. The real reasons are messier and far more preventable.
Vague scope is how projects die. Here's the framework that stops it happening before you've even started.
Every bad engagement had warning signs in the original brief. Here are the 12 we've learned to never ignore.
Clients say they want an AI strategy. What they actually want is someone to make a scary decision feel less scary.