Assistant settings
What the settings on your AI tutor do, and which ones you can leave alone.
Your AI tutor is called an assistant in SLAN. Open it from your course and click Assistants. Most of these settings are fine as they are.
Here is one tutor taken apart. Press a part to see what is inside it.
Market-entry coach
Multi Stage
- 1
Overview
Its name, what it's for, and the conversation starters a student sees when they first open it.
- 2
Stages
A Multi stage assistant walks the student through stages in order. Each has its own instructions, the frameworks it applies, and a criteria of done that tells the assistant when to move on.
- 3
Guardrails
Rules it keeps to in every conversation.
- 4
Final Report
What the student leaves with at the end. Multi stage assistants only.
- 5
Relationships
The course it belongs to, the files and frameworks it draws on (Related Context), and the assistants it can hand a student over to.
- 6
Advanced Parameters
How it tutors (its Tutoring Approach) and the example exchanges that ground it.
- 7
Access Control
Who can use it: the access level, the cohorts it's assigned to, and any widgets that embed it on another site.
AI Model
The AI model is the engine behind your tutor. Open Model Settings in the left menu to see it. Auto (Recommended) is the default. SLAN picks a model for each question. Leave it on Auto unless you have a reason to change it.
You can also pick one model:
- Gemini 2.5 Flash
- GPT-5 Mini
- GPT-4o Mini
- GPT-5 Nano
- Claude Haiku 4.5
If your account keeps requests in the EU, the list is shorter. It shows only Auto, Gemini 2.5 Flash (EU) and Claude Haiku 4.5 (EU).
Temperature and Top-p
Temperature sets how much your tutor's wording varies. Lower is steadier. Higher is more varied. Top-p limits variety in a similar way.
On Auto, both show Managed automatically. Pick a specific model and they show 0.7 and 0.9. GPT-5 Mini and GPT-5 Nano show Not supported by this model. Leave both alone.
Conversation starters
Starters are the buttons students see when they open the tutor. You add between 1 and 5 on Overview. Click Add starter for each one.
Write what a student would really ask:
- "Can we work through a DCF model together?"
- "I have an exam tomorrow. Can we do a quick revision?"
Avoid "Hello" or "Ask me anything". They tell the student nothing.
Access level
This decides who can use your tutor. You change it in Access Control. The rules live on access levels.
Rate limits
Optional. In Model Settings, you can cap the messages one student sends per minute. You can also cap the messages the whole tutor answers per hour. Leave them off if you are unsure.
Performance
Overview shows Total Conversations, Average Rating and Avg Response Time. They appear only after students have used the tutor.