Skip to main content
MIND Intelligence is the conversational brain of the platform. A student types a message (or taps a quick action) and MIND replies in real time, grounded in who the student is and what they’re studying. It is router-first and deterministic by default: every message is classified into a lane before an answer is generated, so behaviour stays predictable and inexpensive. The core is channel-agnostic — the same services that power text today are designed to power voice later.

​
One endpoint

Everything flows through a single streaming endpoint: 
POST /v1/mind/chat
The reply streams back over Server-Sent Events (SSE) — the student sees tokens appear as they are generated, and the conversation is remembered across requests.

​
Quick actions vs. free chat

The client sends an action with every message. When the student taps a quick action it is set explicitly; when they just type, it defaults to free-chat and the router decides the lane.
ActionMeaning
free-chatThe student just typed. The router classifies the message and may upgrade it to one of the lanes below.
explain-conceptExplain a concept, topic, or piece of course material.
submission-guideHelp with the mechanics of submitting assessed work.
advisorThe student wants a human — advice, escalation, or to book a meeting.
Today all four lanes answer through the conversational engine. submission-guide (a guided workflow) and advisor (an escalation to a human) get their own dedicated flows as the authenticated-student profile becomes available.

​
Where to go next

Architecture

The full path of a message: quota, guardrails, routing, the conversational graph, streaming, and resumption.

Streaming

How SSE works, why we chose it over WebSockets, and how the frontend consumes it.

Chat API

The POST /v1/mind/chat reference — request body and the SSE event contract.