DEV Community

kanta13jp1
kanta13jp1

Posted on

Building an AI Provider Encyclopedia with Supabase + Flutter — 93 Providers and Counting

#flutter #supabase #ai #buildinpublic

Building an AI Provider Encyclopedia with Supabase + Flutter

What is "AI University"?

AI University is a learning feature in 自分株式会社 (Jibun Kabushiki Kaisha) that covers 93 AI providers:

Google / OpenAI / Anthropic / Mistral / DeepSeek / Groq /
Nebius / DeepInfra / Fireworks AI / SambaNova / ... (93 total)
Enter fullscreen mode Exit fullscreen mode

For each provider: overview, available models, API usage guide. With quizzes, scores, leaderboards, and badges to make it a game.

The Schema That Scales 

CREATE TABLE ai_university_content (
  id uuid DEFAULT gen_random_uuid() PRIMARY KEY,
  provider text NOT NULL,
  category text NOT NULL,  -- "overview" | "models" | "api" | "news"
  title text NOT NULL,
  content text NOT NULL,   -- Markdown
  published_at date,
  UNIQUE(provider, category)
);
Enter fullscreen mode Exit fullscreen mode

The UNIQUE(provider, category) constraint means each provider has exactly one record per category. ON CONFLICT DO UPDATE makes upserts idempotent — run the same migration twice with no side effects.

DB-Driven Tabs in Flutter 

Future<List<String>> _fetchProviders() async {
  final response = await Supabase.instance.client
      .from('ai_university_content')
      .select('provider')
      .order('provider')
      .limit(200);

  return (response as List)
      .map((r) => r['provider'] as String)
      .toSet()
      .toList();
}
Enter fullscreen mode Exit fullscreen mode

No hardcoded tab list. Add a DB record → tab appears automatically. This is why we could go from 9 to 93 providers without touching UI code.

Adding a Provider: 15 Minutes Per Entry

  1. WebSearch for latest info (models, API, pricing)
  2. Write migration SQL (3 categories: overview, models, api)
  3. Apply: supabase db push
  4. Flutter: nothing to change 
-- Example: adding GMI Cloud
INSERT INTO ai_university_content (provider, category, title, content, published_at)
VALUES
  ('gmi_cloud', 'overview', 'GMI Cloud Overview', '## What is GMI Cloud\n\nGPU cloud...', '2026-04-19'),
  ('gmi_cloud', 'models',   'GMI Cloud Models',   '## Available Models\n\n...', '2026-04-19'),
  ('gmi_cloud', 'api',      'GMI Cloud API',       '## API Usage\n\n...', '2026-04-19')
ON CONFLICT (provider, category) DO UPDATE
  SET content = EXCLUDED.content,
      published_at = EXCLUDED.published_at;
Enter fullscreen mode Exit fullscreen mode

Automatic News Updates (Every 2 Hours) 

# ai-university-update.yml
on:
  schedule:
    - cron: '0 */2 * * *'

jobs:
  update:
    steps:
      - name: Fetch and upsert news
        # Fetches each provider's RSS / blog → upserts news category
Enter fullscreen mode Exit fullscreen mode

93 providers × every 2 hours. The "news" tab always shows current information.

The Learning Flow 

Select provider
  → Read overview / models / api
  → Quiz (3 questions)
  → Score recorded (ai_university_scores)
  → Badge awarded (ai_university_badges)
  → Leaderboard updated
Enter fullscreen mode Exit fullscreen mode

FSRS (spaced repetition algorithm) calculates review due dates. A streak counter shows consecutive learning days to motivate daily habits.

Why This Architecture Works at Scale

Design decision Why it scales
UNIQUE(provider, category) Upserts are safe; no duplicate content
DB-driven tabs Adding provider = adding DB record
Markdown content Rich formatting without schema changes
Auto-update cron Content stays fresh without manual work

Going from 9 to 93 providers required zero UI code changes. The investment in DB-driven architecture paid off at provider #10 and compounded every provider after.

Building in public: https://my-web-app-b67f4.web.app/

Flutter #Supabase #AI #buildinpublic

Top comments (0)