AI agent
Software that takes a goal, plans the steps, uses tools and keeps working until it is done. On the buyer's side it is the procurement of the future; on yours it maintains the plan.
An AI agent is a model given a goal rather than a question: it plans the steps, uses tools such as search, files and other software, checks its own progress and keeps working until the goal is achieved or it gets stuck. The book meets agents in two places, and they are not the same agent.
On the buyer’s side, agents are the machine buying that Gartner expects to intermediate fifteen trillion dollars of B2B spend by 2028, comparing suppliers on criteria a human never sees. On the supplier’s side, an agent on a single laptop is what keeps a living account plan current, reading each new source against the schema and updating the pages it touches. Gartner also names the risk, agent drift, where an agent left unsupervised wanders from the goal it was given. The answer is the same as everywhere else in the book: a schema, and a human who reads what came back.
Set out in full in: Creating Customer Value Propositions with AI (Kogan Page, 2027), chapters 1, 2, 4 and 9.
Related terms
Machine buying
B2B purchasing done by software agents on the buyer's side. Around a third of it by 2030, and the transactional layer of selling with it.
The five ages of AI
Prompt, agent, collaborator, advisor, architect. Where you are, where the work is going, and which end is still informed speculation.
The living account plan
Something closer to a brain than a binder: an account plan an AI maintains from raw sources, against your methodology, so it is current when you open it.
Copilot to autopilot
A copilot suggests, drafts and assists beside the human. An autopilot takes the controls on whole tasks. The line between them is drawn by judgement.
See the method run on one real account in three days: the One Page Proposition.