Project case study

Chatmate

Student mental health services are stretched thin. This platform gives students a private, always available space to talk and knows exactly when the conversation needs a human instead.

Chatmate screenshot

The problem

Students often need someone to talk to at 2am, long before a counsellor is free, and a chatbot that misses a crisis is worse than having none at all.

What I built

  • 01Chatmate, a warm AI companion on Gemini that reflects feelings back before it offers any perspective.
  • 02Replies stream in token by token, with automatic fallback to another model when one is rate-limited.
  • 03Crisis handling: talk of self-harm or danger triggers a calm response and real Nigerian helplines (MANI, 24/7).
  • 04One-tap escalation that hands the conversation to a human counsellor.
  • 05A counsellor dashboard with pending and resolved cases, resolution notes and average response time.
  • 06Conversation export to PDF, so a student can take the conversation to a professional.
  • 07No accounts needed to start talking.

How it works

  1. 1

    Talk

    Student opens the chat, with no forms in the way.

  2. 2

    Listen

    Gemini answers using a carefully written listening-first system prompt, streamed live.

  3. 3

    Detect

    Signs of a crisis trigger safe wording and helplines.

  4. 4

    Escalate

    Student hands the conversation to a human in one tap.

  5. 5

    Resolve

    A counsellor picks up the case from the dashboard and records the outcome.