The Pub Agent blog
Notes from the Frontier.
What AI actually does for hospitality businesses: what to adopt, what to ignore, and how to keep the human bit human. Written by me, from real work. Subscribe by RSS.
Closest to the money
In 1992 Steve Jobs quoted the Pentagon's IT chief on where computer money pays back: operations, not management. Pubs are about to learn the same lesson with AI, and this time the tools can actually reach the bar (part one of two).
Lean on humans, strong on process
Part two: the money for AI goes closest to the card tap, and the floor is as close as it gets. Table 42, the jazz standard, the gear change at ten to five, and why it is nearly always the process that fails, not the people.
Signal, not noise
Steve Jobs and Musk get credited with vision, but their sharpest edge was refusal: they filtered noise better than anyone, and it made them hard to work for. In part three of three, AI can now do the filtering, so focus no longer has to cost you your manners.
Increase the signal, reduce the noise = £
Beacon turns an estate's digital noise into signals leaders can act on. What live work across four pubs could become across 240, from review health and listing drift to brand risk and campaign reach.
Why we lose the AI answers
AI answers reward one clear claim, a citable source and numbers that back it. Evidence from three pub pilots shows exactly why each loses recommendations, and which fix to test first.
WWW used to mean the web. Now it means What, Why, Where.
Google's AI now answers most of what a guest wants to know before they ever open your website. Here's the one part of every search question it still can't write for you.
The octopus or the ammonite: an operator's read on AI in pubs
Stephen Wunker and Jonathan Brill's octopus model, read by the person who runs both the tech stack and the cellar. Push intelligence to the arms, wire the nervous system, or go the way of the well-armoured shellfish that went extinct.
Own the harness: local AI in the ERP year
When a medium-sized business moves from SAP into Microsoft, the AI decision is not which model to rent. It is whether the company owns the data, tools and guardrails around the model.
Harnesses: the pub operating system hiding in plain sight
If you care about AI systems, you will have heard the word harness. Hospitality should care too, because a good pub has always run on one.
Agentification: when AI replaces process, humans replace it with revenue
AI agents should absorb the admin, not the people. The operating rule that makes AI adoption work in hospitality, and the three-wave rollout that builds trust inside the business before it touches the guest.
Cheap AI is a trap if your data is trapped
The price of AI is falling fast, yet AI bills keep going up. The moat is not the model. It is your context, and whether a machine can actually reach it.
From SEO to GEO: the new front door
Guests now ask AI where to eat and drink before they ever see your website. SEO is about being found by humans. GEO is about being understood by machines. Here is what that changes.