How to Spot When You're Negotiating Against Someone Using AI

AI-augmented negotiation is already happening. In boardrooms, procurement functions, and commercial teams at leading organisations, the preparation that used to take days now takes hours — and it is more rigorous than anything a human team produced before.

If you are sitting across the table from a counterparty and you cannot tell whether they are AI-augmented, you are operating blind. Here is what to look for.

The Preparation Tells

The first signal is the quality and structure of their opening position. AI-prepared negotiators tend to arrive with unusually precise anchors. Not round numbers — specific ones. Not broad objectives — a tiered set with explicit fallback logic.

When someone opens with a position that appears to have accounted for three levels of your potential response, they did not arrive at that position through instinct. It was modelled.

The Linguistic Patterns

AI-assisted communication has characteristic patterns that are learnable.

Watch for unusual structural consistency. Emails and proposals that are AI-drafted tend to have parallel sentence structure, balanced paragraph lengths, and a certain grammatical evenness that human writing lacks. There is a smoothness that comes from language model generation — clauses that are individually well-formed but collectively slightly too tidy.

In conversation, watch for over-prepared transitions. The counterparty who responds to your unexpected point with a clean, well-structured pivot did not spontaneously organise that response. They are working from a prepared scenario that included your point as a possibility.

The Behavioural Baseline

The more reliable signal is deviation from baseline.

In any negotiation, a counterparty establishes a behavioural baseline in the early sessions — a typical response speed, a characteristic level of ambiguity in their language, a pattern of when they use questions versus statements.

AI-prepared negotiators tend to be more consistent than humans and then suddenly less so. The preparation covers the anticipated scenarios with precision. When you move into unexpected territory — a concession they did not model, a threat they did not anticipate — watch for a reset. A slightly longer pause. A request to take the point offline. A return to structured language after a moment of genuine uncertainty.

That is the edge of their preparation. That is where you push.

What To Do With This Information

The goal is not to prove they are using AI. The goal is to identify the boundaries of their preparation and operate at those edges.

AI preparation is comprehensive within the scenarios that were modelled. It is brittle at the margins. If you can identify what they modelled for and what they did not, you know where their composure is real and where it is borrowed.

The counterparty who is perfectly prepared for everything you planned to do, but visibly recalibrating when you deviate — that is your opening.