Summary

Cloudflare AI Search now exposes OpenAI-compatible REST endpoints for search and chat. The endpoints accept a messages array (roles and content), allowing you to maintain session context across turns and reuse existing OpenAI SDKs and tooling.

Key Points

• Endpoints: POST /accounts/{account_id}/ai-search/instances/{name}/chat/completions and POST /accounts/{account_id}/ai-search/instances/{name}/search.
• Request format: use a messages array with entries like {"role":"system","content":"..."} and {"role":"user","content":"..."} to pass prior turns and preserve context.
• Example cURL: curl https://api.cloudflare.com/client/v4/accounts/{ACCOUNT_ID}/ai-search/instances/{NAME}/chat/completions -H "Content-Type: application/json" -H "Authorization: Bearer {API_TOKEN}" -d '{ "messages": [ { "role": "system", "content": "You are a helpful documentation assistant." }, { "role": "user", "content": "How do I get started?" } ] }'
• Migration: if you use the older AutoRAG endpoints (/autorag/rags/), migrate to the new API as soon as possible — new features will be added only to the new endpoints. Existing AutoRAG endpoints will continue to work for now; Cloudflare will provide at least 90 days notice before changes.
• Practical checklist: update client endpoint URLs, switch payloads to the messages array, preserve and replay prior messages for multi-turn context, verify Authorization: Bearer {API_TOKEN} and run integration tests.