Skip to main content
SENTi-radar integrates with multiple third-party APIs to enhance data collection and AI analysis. This guide walks you through obtaining API keys for each service.
All AI API keys listed here are optional. SENTi-radar has built-in fallback mechanisms and works without any external AI keys by using local keyword-based analysis.

​
Overview

ServicePurposeRequiredFree TierMonthly Quota
Scrape.doX & Reddit scraping⭐ Recommended✅ Yes1,000 requests
YouTube Data APIVideo commentsOptional✅ Yes10,000 units/day
OpenAIAI insights (Tier 1)Optional❌ Pay-per-use$0.15/1M input tokens
Google GeminiAI analysis (Tier 2)Optional✅ Yes15 RPM (free tier)
GroqLLM fallback (Tier 3)Optional✅ Yes14,400 requests/day
Parallel.aiSocial search fallbackOptional❌ Paid onlyVaries by plan
This is the primary data source for X (Twitter) and Reddit. Without it, the app falls back to YouTube → Parallel.ai → Algorithmic generation.
1

Create account

  1. Go to scrape.do
  2. Click Sign Up or Start Free Trial
  3. Verify your email address
2

Get API token

  1. Log in to Scrape.do dashboard
  2. Navigate to API Tokens section
  3. Copy your default token or create a new one
  4. Note your free tier credits (usually 1,000 requests)
3

Add to environment

Client (.env):
VITE_SCRAPE_TOKEN=your-scrape-do-token-here
Server (Supabase):
supabase secrets set SCRAPE_DO_TOKEN=your-scrape-do-token-here
Pricing:
  • Free tier: 1,000 requests
  • Standard request: 1-5 credits
  • With rendering: 5-10 credits
  • Residential proxies: 25-50 credits
Documentation: docs.scrape.do

​
YouTube Data API v3

Enables fetching video search results and comments for sentiment analysis.
1

Create Google Cloud project

  1. Go to Google Cloud Console
  2. Click Select a project → New Project
  3. Enter project name: senti-radar-youtube
  4. Click Create
2

Enable YouTube Data API

  1. In the Cloud Console, go to APIs & Services → Library
  2. Search for “YouTube Data API v3”
  3. Click YouTube Data API v3
  4. Click Enable
3

Create API credentials

  1. Go to APIs & Services → Credentials
  2. Click Create Credentials → API Key
  3. Copy the generated API key
  4. (Optional) Click Restrict Key to add application restrictions:
    • Application restrictions: HTTP referrers or IP addresses
    • API restrictions: Restrict to YouTube Data API v3
4

Add to environment

Client (.env):
VITE_YOUTUBE_API_KEY=AIzaSyDxxxxxxxxxxxxxxxxxxxxxxxxxxx
Server (Supabase):
supabase secrets set YOUTUBE_API_KEY=AIzaSyDxxxxxxxxxxxxxxxxxxxxxxxxxxx
Quota Limits:
  • Default: 10,000 units per day
  • Search request: 100 units
  • Video details: 1 unit
  • Comments: 1 unit
Cost: Free tier is generous. Paid plans available if you exceed quota. Documentation: YouTube Data API Docs
YouTube API quota resets daily at midnight Pacific Time (PT). Plan your requests accordingly.

​
OpenAI (GPT-4o mini)

Tier 1 AI provider for strategic insights and chat. Provides the highest quality analysis.
1

Create OpenAI account

  1. Go to platform.openai.com
  2. Click Sign up and create an account
  3. Complete email verification
2

Add billing

  1. Navigate to Billing Settings
  2. Add a payment method
  3. Set a usage limit to control costs (recommended: $10/month)
3

Create API key

  1. Go to API Keys
  2. Click Create new secret key
  3. Name it: senti-radar-production
  4. Copy the key (starts with sk-proj-)
Copy this key now! You won’t be able to see it again.
4

Add to environment

Client (.env):
VITE_OPENAI_API_KEY=sk-proj-xxxxxxxxxxxxxxxxxxxxxxxxxx
5

Verify integration

Test the API key:
curl https://api.openai.com/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -d '{"model": "gpt-4o-mini", "messages": [{"role": "user", "content": "Test"}]}'
Model Used: gpt-4o-mini Rate Limits:
  • Free tier: Not available (pay-per-use only)
  • Tier 1: 500 RPM, 200,000 TPM
  • Tier 2+: Higher limits based on usage
Pricing:
  • Input: $0.150 per 1M tokens
  • Output: $0.600 per 1M tokens
  • Typical insight report: ~$0.001-0.003 per analysis
Fallback: If OpenAI fails → Gemini → Groq → Local analysis Documentation: OpenAI API Docs Used in:
  • Client-side AI Insights Panel (strategic reports and chat)

​
Google Gemini AI

Tier 2 AI provider for sentiment analysis, emotion detection, and summary generation.
1

Get API key

  1. Go to Google AI Studio
  2. Click Get API key
  3. Sign in with your Google account
  4. Click Create API key
  5. Select a Google Cloud project (or create new)
  6. Copy the generated API key
2

Add to environment

Client (.env):
VITE_GEMINI_API_KEY=AIzaSyCxxxxxxxxxxxxxxxxxxxxxxxx
Server (Supabase):
supabase secrets set GEMINI_API_KEY=AIzaSyCxxxxxxxxxxxxxxxxxxxxxxxx
3

Verify integration

Test the API key:
curl "https://generativelanguage.googleapis.com/v1beta/models/gemini-2.0-flash:generateContent?key=YOUR_API_KEY" \
  -H 'Content-Type: application/json' \
  -d '{"contents":[{"parts":[{"text":"Test"}]}]}'
Model Used: gemini-2.0-flash Rate Limits (Free Tier):
  • 15 requests per minute (RPM)
  • 1 million tokens per minute (TPM)
  • 1,500 requests per day (RPD)
Fallback: If Gemini fails or is unavailable, SENTi-radar uses a local keyword-based sentiment engine. Pricing:
  • Free tier: Generous limits for development
  • Pay-as-you-go: $0.075 per 1M input tokens
Documentation: Gemini API Docs Used in these functions:
  • analyze-sentiment (sentiment + emotion classification)
  • generate-insights (AI summaries and key takeaways)
  • analyze-topic (comprehensive topic analysis)

​
Groq

Secondary LLM option for streaming AI summaries (alternative to Gemini).
1

Create Groq account

  1. Go to console.groq.com
  2. Sign up with email or Google account
  3. Verify your email
2

Generate API key

  1. Log in to Groq Console
  2. Click Create API Key
  3. Enter a name: senti-radar-production
  4. Click Create
  5. Copy the API key (starts with gsk_)
3

Add to environment

Client (.env):
VITE_GROQ_API_KEY=gsk_xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
Groq is primarily used client-side. No need to set as Supabase secret unless you modify edge functions to use Groq.
Models Available:
  • llama-3.1-70b-versatile
  • llama-3.1-8b-instant
  • mixtral-8x7b-32768
  • gemma2-9b-it
Rate Limits (Free Tier):
  • 14,400 requests per day
  • 30 requests per minute
  • Very fast inference speed (100+ tokens/sec)
Pricing:
  • Free tier: 14,400 requests/day
  • Pay-as-you-go: $0.10 per 1M tokens (Llama 3.1 8B)
Documentation: Groq Docs

​
Parallel.ai (Optional)

Social search fallback when Scrape.do is unavailable.
Parallel.ai is a paid service with no free tier. Only configure if you have a paid account and need additional fallback beyond YouTube.
1

Sign up for Parallel.ai

  1. Go to parallel.ai
  2. Contact sales or sign up for a plan
  3. Complete onboarding and payment setup
2

Get API key

  1. Log in to Parallel.ai dashboard
  2. Navigate to Settings → API Keys
  3. Generate a new API key
  4. Copy the key
3

Add to Supabase secrets

supabase secrets set PARALLEL_API_KEY=your-parallel-api-key
Parallel.ai is only used server-side in the fetch-twitter edge function. No client-side configuration needed.
Used in: fetch-twitter edge function (Step 2 fallback) Documentation: Contact Parallel.ai support for API docs

​
Verification Checklist

After obtaining your API keys, verify they’re configured correctly:
1

Check environment files

# View your .env file (client-side)
cat .env

# Should contain:
# VITE_SUPABASE_URL=...
# VITE_SUPABASE_PUBLISHABLE_KEY=...
# VITE_SCRAPE_TOKEN=...
# VITE_YOUTUBE_API_KEY=...
# VITE_GEMINI_API_KEY=...
# VITE_GROQ_API_KEY=...
2

Check Supabase secrets

# List all edge function secrets
supabase secrets list

# Should show:
# SCRAPE_DO_TOKEN
# YOUTUBE_API_KEY
# GEMINI_API_KEY
# PARALLEL_API_KEY (optional)
# SUPABASE_URL
# SUPABASE_SERVICE_ROLE_KEY
3

Test API connections

Start the development server:
npm run dev
  1. Create a new topic in the UI
  2. Watch browser console for API calls
  3. Check for successful responses (HTTP 200)
  4. Verify posts are fetched and analyzed
4

Test edge functions

# Test fetch-twitter function
supabase functions serve fetch-twitter

# In another terminal, invoke the function
curl -i --location --request POST \
  'http://localhost:54321/functions/v1/fetch-twitter' \
  --header 'Content-Type: application/json' \
  --data '{"topic_id":"test-uuid-here"}'

​
Best Practices

​
Security

  • Never commit API keys to Git
  • Use .env files (already in .gitignore)
  • Rotate keys regularly
  • Use environment-specific keys (dev vs. prod)

​
Key Rotation

# Rotate client keys
1. Update .env with new key
2. Restart dev server: npm run dev

# Rotate server keys
1. Update Supabase secrets: supabase secrets set KEY=new-value
2. Redeploy functions: supabase functions deploy

​
Rate Limiting

  1. Implement exponential backoff for rate limit errors (429)
  2. Cache API responses to reduce duplicate requests
  3. Monitor quota usage in each service’s dashboard
  4. Set up alerts when approaching quota limits

​
Cost Optimization

  1. Use free tiers first before upgrading to paid plans
  2. Enable only necessary APIs (start with Scrape.do + Gemini)
  3. Monitor usage in each provider’s dashboard
  4. Implement caching to reduce redundant API calls
  5. Use fallbacks wisely:
    • Scrape.do (primary)
    • YouTube (secondary)
    • Parallel.ai (tertiary, paid)
    • Algorithmic (guaranteed fallback)

​
Troubleshooting

​
API Key Not Working

Problem: API returns 401/403 errors Solutions:
  1. Verify key is copied correctly (no extra spaces)
  2. Check key is active in provider dashboard
  3. Verify API is enabled (for Google APIs)
  4. Check IP restrictions (if configured)
  5. Restart dev server after adding keys

​
Quota Exceeded

Problem: API returns 429 or quota error Solutions:
  1. Check usage in provider dashboard
  2. Wait for quota reset (usually daily)
  3. Implement request caching
  4. Upgrade to paid tier
  5. Use fallback providers

​
Edge Function Can’t Access Secrets

Problem: Function logs show “undefined” for secrets Solutions:
  1. Verify secrets are set: supabase secrets list
  2. Redeploy function after setting secrets
  3. Check secret name matches exactly (case-sensitive)
  4. View function logs: supabase functions logs <function-name>

​
YouTube API Quota Depletes Quickly

Problem: Hitting 10,000 unit daily limit Solutions:
  1. Each search costs 100 units (limit to 100 searches/day)
  2. Implement result caching in database
  3. Reduce maxResults parameter (default 15)
  4. Request quota increase in Google Cloud Console
  5. Use YouTube only as fallback (not primary source)

​
API Key Reference Summary

# Required
VITE_SUPABASE_URL=https://your-project.supabase.co
VITE_SUPABASE_PUBLISHABLE_KEY=eyJhbGciOi...

# Recommended
VITE_SCRAPE_TOKEN=abc123...

# Optional
VITE_YOUTUBE_API_KEY=AIzaSyD...
VITE_GEMINI_API_KEY=AIzaSyC...
VITE_GROQ_API_KEY=gsk_...

​
Next Steps

Environment Variables

Complete environment configuration guide

Supabase Setup

Deploy edge functions and database

Scrape.do Integration

Advanced scraping configuration

Deployment

Deploy your app to production

Build docs developers (and LLMs) love

Get started for freeTalk to us