Peer is building the behavioural dataset and evaluation benchmark for conversational AI operating in commercial environments.

As AI agents increasingly participate in sales, customer support, and commercial conversations, the industry lacks reliable training data and evaluation standards for the behaviours required to succeed in these interactions.

Peer addresses this gap through two core components:

The Peer Sales Behaviour Dataset
A structured dataset containing over 250,000 annotated behavioural examples extracted from 10,000 real commercial conversations, organised across 150 defined sales skills. Each example demonstrates a specific behaviour such as discovery questioning, objection handling, value articulation, or negotiation, and is scored using a consistent expert rubric.

The Peer Behaviour Benchmark
A framework for evaluating how effectively AI agents perform in commercial conversations. AI companies can test agent responses against real behavioural scenarios and receive structured performance scores across key commercial competencies.

Together, these form the training and evaluation layer for conversational AI in commercial environments.

Peer works with AI companies, research teams, and enterprise AI groups building conversational agents that interact with customers and prospects.

If you are building AI systems that participate in commercial conversations, Peer provides the behavioural ground truth needed to train and evaluate them.

Peer

Org chart

Teams

Offices