KIOKU™ is a production-ready vector memory API. Store, search, and share memories across agents, sessions, and models.
Production-grade memory infrastructure — no DIY vector DB setup required.
70% vector cosine similarity + 30% full-text search, weighted by importance score. Always finds the most relevant memory.
GPT-4o-mini automatically decides: Add, Update, Delete, or Noop. No duplicate memories — ever.
Frequently accessed memories served from Redis in <50ms. Automatic TTL and cache invalidation on write.
Full namespace isolation: tenant → project → agent → session. Perfect for SaaS AI products.
Multi-agent consensus sessions. Multiple agents collaborate on a decision, with persistent outcome storage.
Per-operation metering: writes, reads, searches, deliberations. Full cost breakdown via /v1/usage.
From zero to persistent memory in under 2 minutes.
# 1. Write a memory curl -X POST https://usekioku.com/v1/memories \ -H "Authorization: Bearer YOUR_API_KEY" \ -H "Content-Type: application/json" \ -d '{"content":"User prefers dark mode and concise responses","type":"semantic","importance":0.9}' # 2. Search memories curl -X POST https://usekioku.com/v1/memories/search \ -H "Authorization: Bearer YOUR_API_KEY" \ -d '{"query":"user preferences","top_k":5}' # 3. Check usage curl https://usekioku.com/v1/usage \ -H "Authorization: Bearer YOUR_API_KEY"
Start free. Scale when you need to.
Per-operation pricing: Write $0.002 · Read $0.001 · Search $0.0015 · Deliberation $0.05
Free to start. No credit card required.