All answers
How it works

How do I add relationship context to an AI sales agent?

Give an AI sales agent structured information about relevant people, connection evidence and permitted next actions, rather than a flat contact list. Include source dates, confidence and connector confirmation. Use relationship context to support research and routing, while keeping approval for introductions and external messages separate from the agent's ability to read the data.
October 8, 2026

Model the necessary context

Represent people, organisations, current and past roles, relationship types and evidence. Include who owns the relationship and whether anyone has confirmed familiarity or willingness for the request.

Retrieve around the task

A request to reach a buyer should return relevant stakeholders and possible paths, not every connection in the organisation. Keep sensitive fields out unless needed and authorised.

Separate inference from action

The agent can propose a former-colleague route. A person should validate it and decide whether to make an introduction. Do not turn a shared employer into an automated endorsement.

Test the result

Use known identities and relationships to check false matches, stale records and inappropriate routing. Assess whether the agent improves completed work and commercial relevance.

Orbb's relationship-data direction is relevant to this architecture. Confirm available APIs, MCP tools and integration scope directly rather than assuming a production interface or permission model from positioning material.

Related questions

Orbb researches the relationships your company already has — across customers, employees, investors and partners — then runs the introduction end to end, from finding the path to the meeting in the calendar.
See it on your own accounts