Mentor with Memory

Configure an optional journey-aware mentor using the Markdown guidance developed with your team.

A mentor aligned to your process

Mentor with Memory is an optional AI agent role and is not available in every Talvi installation. Where this capability is provided, Talvi will have worked with your team to develop the agent and its journey guidance around your organisation’s needs. Use the journey document agreed for your implementation.

A Mentor with Memory agent uses a special Markdown journey document to understand the flow it should support. Once the end user selects that agent, the document guides how it communicates: the stages that matter, useful questions, relevant next steps, and boundaries it should respect.

What the journey can describe

Example	How the mentor could help
Employee onboarding	Guide a new starter through orientation, finding resources, and preparing for a first team meeting, taking account of what they have already shared.
A learning opportunity	Understand the learner’s goal and experience, suggest relevant material, and help them reflect before moving to the next topic.
A workplace process	Explain a sequence such as preparing a request, gathering information, reviewing it with a manager, and submitting it through the correct channel.

These are illustrative uses. Your team and Talvi define the actual stages, wording, resources, and boundaries for your agent.

How the guidance and memory work together

The journey document gives the agent a process to draw on. Its Mission defines its purpose, and its Dataset supplies the knowledge it can retrieve to support factual answers. Conversation history and available memory help it use relevant details the person has already provided, rather than starting every exchange from scratch.

The agent can ask a focused clarification when it needs more context and offer relevant follow-up questions or next steps. The user can change direction: the journey guides the conversation without requiring every person to follow an identical script.

Memory availability and continuity depend on your implementation and the surface where the agent is used. A saved journey document is shared guidance for the agent; it is not a record of an individual’s progress. Do not treat conversational guidance as proof that an employee has completed a task or that a business system has been updated.

Prepare the Markdown journey file

Markdown is a text format that uses headings and lists to organise information. Talvi works with your team to turn the intended journey into a clear document. It may describe:

The audience, goal, and limits of the mentor’s role.
Each stage, its purpose, and any prerequisites.
Questions that help establish what the person needs.
Clear descriptive steps and signs that the person is ready to move on.
Approved supporting resources and when to involve a colleague or specialist.
Communication guidelines, including what the agent should ask rather than assume.

For example, a small section of an onboarding journey might look like this:

# New starter onboarding journey

## Purpose
Help a new starter understand the onboarding process and find help.
Ask what they have already completed before suggesting a next step.

## Stage 1: Get oriented
Goal: understand the team's role and where to find approved resources.

Suggested guidance:
1. Ask which team the person has joined, if it is not already known.
2. Help them find the team introduction in the available knowledge.
3. Ask whether they need help preparing for their welcome meeting.

## Stage 2: Prepare for the welcome meeting
Goal: arrive with a short list of questions and any access blockers.

Suggested guidance:
1. Help the person list what they want to understand about their role.
2. Suggest raising unresolved access issues with the designated team.
3. Let the person confirm what they have done; do not assume completion.

## Communication guidelines
Answer the person's current question before suggesting another step.
Use approved source material for factual details.
Ask for clarification when a required detail is missing.

This is a generic illustration, not a required template or a complete deployment journey. Your production document should use the process developed with Talvi and your team.

Upload it in AI Agent Studio

Open More.. → Super Admin → AI Agent Studio and start creating an agent.
Under Role, select Mentor with Memory, when available.
In the Mission area, complete the agent’s purpose. Selecting this role also reveals Mentor journey document.
Use the file input to upload the agreed Markdown file. The picker accepts .md, .markdown, and .txt; use a Markdown file for your journey.
Check the displayed filename and the text loaded into the editor. You can also paste the Markdown directly into that editor.
Choose the Dataset, optional Skills, and Output, then complete Save AI Agent. Studio requires non-empty journey text for this role before saving.

Uploading the file loads its text into the creation form. Finish saving the agent to retain it. The journey input is separate from a dataset upload: use Create Dataset or Data Import for the documents that supply the agent’s knowledge.

For a website output, also complete the embedded bot configuration and credentials.

Update an existing mentor

Open the saved agent from AI Agent Studio → Custom AI Agents. Agents with this role show a Mentor journey document editor, including in the embedded-agent workspace.

Upload a revised file or edit the Markdown in the text area. Review the changes and select Save Journey. Uploading a replacement or typing into the editor alone does not save the update. Use Cancel to discard unsaved edits.

Coordinate substantial process changes with your team and Talvi. The updated document changes the guidance this agent uses in subsequent conversations; it does not rewrite earlier answers or confirm that a user has completed the revised process.

Check the experience

Select the mentor agent and try a few representative situations: a person just starting, someone partway through, and someone asking a question outside the usual sequence. Check that the guidance is appropriate, uses the dataset for factual detail, and respects corrections from the user.

Review relevant conversations through the agent’s chat history and reports. If the role or journey input is missing, check with your administrator or Talvi contact whether this capability is included in your installation.