Intro to AI Tutor
An AI Tutor is a conversational agent that teaches from your own documents. Your team joins a live session with the tutor and works through the material by talking about it, instead of reading a deck or watching a recording.
The difference between a tutor and a roleplay agent is who is playing which part. In a roleplay, the agent plays a buyer or a customer and your team practises selling or supporting. With a tutor, the agent plays the teacher — it knows the playbook, the product doc, or the call guide you attached, and it walks the learner through it, asks questions, and corrects them.
You'll find AI Tutor in the left sidebar under Roleplay.
The AI Tutor list.
What a tutor is built from
Every tutor has three parts.
Resources. The documents the tutor teaches from — playbooks, product documentation, call guides, onboarding material. This is the tutor's source of truth, and at least one resource is required. The tutor answers from what you attach, so the quality of the resource is the quality of the session.
A type. What kind of learning the tutor is for:
- Concept — understanding something. Product knowledge, a pricing model, a compliance rule, the reasoning behind a methodology.
- Skill — doing something. Running a discovery call, handling a pricing objection, structuring a demo.
- Tool — using a specific system or application. This type is available on request; contact support to have it enabled for your workspace.
A persona. The tutor's name, gender, language, avatar, voice, and the greeting line it opens the session with. This is what makes the tutor feel like a person rather than a search box.
How your team practises
Tutors run in Voice Mode — a spoken conversation, the same session experience as a voice roleplay. Chat Mode and Video Mode are available on request; contact support to have them enabled for your workspace.
You can also add Roleplay Instructions to a tutor. These are shown to the learner just before the session starts, so you can point them at what to focus on — for example, "Focus on discovery questions before pitching the product."
When to use a tutor instead of a roleplay
Use a tutor when the learner needs to understand something before they can practise it. A new hire who has never seen your pricing model will not get much from a pricing objection roleplay — they will get more from a Concept tutor built on the pricing doc, followed by the roleplay.
A common pattern is to pair them: a tutor to teach the material, then a roleplay agent to pressure-test whether it stuck, then a scorecard to measure the result.
Who can create tutors
Creating and editing tutors is limited to admins and managers. Everyone else can see the tutor list and practise with the tutors that exist, but the Create AI Tutor button will be unavailable.