Test whether AI models mention your brand when answering relevant queries. Runs prompts through GPT-5.4, Claude Sonnet 4, Gemini 3 Flash, Llama 4, and Mistral — all through Cloudflare AI Gateway. No API keys needed.

• Multi-site monitoring — Add multiple domains, each with their own prompts and model configs
• 5 AI models — OpenAI GPT-5.4 Nano, Anthropic Claude Sonnet 4, Google Gemini 3 Flash, Meta Llama 4 Scout, Mistral Small 3.1
• AI prompt generation — Workers AI analyzes your site and suggests relevant test prompts
• Parallel execution — Cloudflare Queues fan out model × prompt jobs for fast results
• Real-time progress — Poll-based UI shows results as they complete
• CSV export — Download results filtered by model or prompt
• SSR — React Router 7 + Hono on Cloudflare Workers, server-rendered

1. Cloudflare account — Sign up
2. Unified Billing credits — Third-party models (OpenAI, Anthropic, Google) are billed through AI Gateway Unified Billing. Load credits in the Cloudflare dashboard.
3. Node.js >= 22 — Required for Wrangler

Workers AI models (@cf/ prefix) use standard Workers AI pricing and do not require Unified Billing credits.

git clone https://github.com/cloudflare/ai-brand-visibility-template
cd ai-brand-visibility-template
npm install

npx wrangler login
npx wrangler kv namespace create AEO_KV       # Replace the example ID in wrangler.jsonc
npx wrangler queues create brand-visibility-jobs

npm run deploy

npm run dev      # Local dev server with hot reload
npm run build    # Production build
npm run deploy   # Build + deploy to Cloudflare

workers/
  app.ts          — Hono entry: API routes + React Router SSR + Queue consumer
  api.ts          — All /api/* endpoints
  queue.ts        — Queue consumer for parallel model inference
app/
  routes/
    layout.tsx    — Dashboard shell (header + sidebar)
    results.tsx   — Results + filters + pagination (index page)
    prompts.tsx   — Prompt management + AI generation
    models.tsx    — Model selection per site
    setup.tsx     — Wizard: site → competitors → prompts → models
  components/     — UI components (Cloudflare dashboard style)
src/
  config.ts       — Model definitions, settings

All models run through env.AI.run() with AI Gateway. No provider API keys required.

Each test runs 5 models × N prompts. With 5 prompts, that's 25 inference calls per test.

• GPT-5.4 Nano — Cheapest OpenAI option, optimized for edge
• Claude Sonnet 4 — Mid-tier Anthropic pricing
• Gemini 3 Flash — Google's fast/cheap option
• Workers AI models — Usage-based, see pricing

Check AI Gateway pricing for current rates.

MIT