GenFit uses Cohere’s Large Language Model (LLM) to power the chatbot on the Challenges page. The bot gives smart, context-aware help so users can join, follow, and complete fitness challenges with less friction.

The bot can:

• Explain rules and requirements
• Show details of active challenges
• Answer questions about goals, metrics, dates and rewards

Users update progress with simple sentences:

I just ran 5 kilometres for the marathon challenge.
Add 45 minutes of cycling to my cardio challenge.

The LLM detects intent and calls the update_progress API with the right fields.

• Show the user’s rank and gap to next place
• Compare the user with friends or the full group
• Give quick stats (avg. pace, total time, etc.)

Using past activity, the bot recommends:

• Challenges that fit the user’s goals
• Complementary challenges that build on current ones
• Local events (if the user shares location)

We call Cohere from the web app with their JavaScript SDK:

import Cohere from 'cohere-ai';

Cohere.init({ apiKey: COHERE_API_KEY });

The API key lives in secure server storage and is sent only from the backend over HTTPS.

Tip: Stay below 75 % of the max context (≈24 k tokens) to keep latency low.

{
  "model": "command-r-plus",
  "max_tokens": 200,
  "temperature": 0.3,
  "p": 0.95,
  "frequency_penalty": 0,
  "presence_penalty": 0
}

• max_tokens – size of the answer
• temperature / p – creativity vs. control
• penalties – reduce repetition

We pass the last 10 chat turns to Cohere (messages array). Older turns are summarised when the total token count nears the limit.

Cohere’s free tier allows ~90 requests / minute. Higher tiers raise this cap.

• Ephemeral chat logs – stored only for the session
• GDPR/CCPA compliant – fitness data kept inside GenFit
• No personal data sent for model training

• Only coaches can change challenge settings
• The bot may ask for clarifications on vague requests
• Voice control is still experimental

• Full voice control
• AI-based workout plans
• Multi-language chat
• Deeper analytics and weekly reports

Learn more: Cohere API docs → https://docs.cohere.com/reference/about

These unit tests make sure that only coaches can create challenges, important fields are present, business rules hold, and the API answers with the right status codes.

1. Authorisation Tests
2. Data Validation Tests
3. Business-Logic Tests
4. Location Tests
5. Edge-Case Tests
6. Implementation Example
7. Running the Tests
8. Best Practices
9. CI Integration

Checks that missing required fields return 400 Bad Request with field errors.

Valid payload returns 201 Created and DB row equals request.

Valid coordinates + name → challenge saved with location data.

Lat outside ±90 or lon outside ±180 → 400 with error.

from django.test import TestCase
from django.urls import reverse
from django.contrib.auth import get_user_model
from rest_framework.test import APIClient
from rest_framework import status
from ..models import Challenge
from datetime import datetime, timedelta
import json

User = get_user_model()

class ChallengeCreationTests(TestCase):
    def setUp(self):
        self.coach = User.objects.create_user(
            username="coach_test",
            email="[email protected]",
            password="securepassword",
            user_type="Coach"
        )
        self.regular = User.objects.create_user(
            username="regular_test",
            email="[email protected]",
            password="securepassword",
            user_type="Regular"
        )
        self.client = APIClient()
        self.url = reverse("create_challenge")
        self.payload = {
            "title": "Test Challenge",
            "description": "This is a test challenge",
            "challenge_type": "Running",
            "target_value": 100,
            "unit": "km",
            "start_date": (datetime.now() + timedelta(days=1)).isoformat(),
            "end_date": (datetime.now() + timedelta(days=30)).isoformat(),
            "min_age": 18,
            "max_age": 65,
            "is_public": True,
            "latitude": 41.0082,
            "longitude": 28.9784,
            "location_name": "Istanbul"
        }

    def test_coach_can_create_challenge(self):
        self.client.force_authenticate(self.coach)
        res = self.client.post(self.url, json.dumps(self.payload), "application/json")
        self.assertEqual(res.status_code, status.HTTP_201_CREATED)
        self.assertTrue(Challenge.objects.filter(title="Test Challenge").exists())

    def test_non_coach_cannot_create_challenge(self):
        self.client.force_authenticate(self.regular)
        res = self.client.post(self.url, json.dumps(self.payload), "application/json")
        self.assertEqual(res.status_code, status.HTTP_403_FORBIDDEN)