Persistent memory in 3 API calls. Store, search, and recall—with Ebbinghaus decay scoring that lets memories fade naturally, just like the real thing.
No SDKs to install. No schemas to define. Just REST.
curl -X POST https://api.smara.io/v1/memories \ -H "Authorization: Bearer smara_your_key" \ -H "Content-Type: application/json" \ -d '{ "userId": "user_42", "content": "Prefers dark mode. Uses vim keybindings.", "tags": ["preferences", "editor"] }'
{ "id": "mem_8f3a...", "decayScore": 1.0, "status": "stored" }
curl https://api.smara.io/v1/memories/search \ -H "Authorization: Bearer smara_your_key" \ -G \ -d userId=user_42 \ -d query="editor preferences" \ -d limit=5
{ "memories": [
{ "content": "Prefers dark mode. Uses vim keybindings.",
"similarity": 0.94, "decayScore": 0.97 }
] }
curl https://api.smara.io/v1/users/user_42/context \
-H "Authorization: Bearer smara_your_key"
{ "userId": "user_42",
"memoryCount": 23,
"context": [
{ "content": "Prefers dark mode. Uses vim keybindings.", "score": 0.96 },
{ "content": "Working on a Rust CLI tool for log parsing", "score": 0.89 }
] }
Memories lose weight over time using a decay curve modeled on human forgetting. Recent memories surface first. Old, unreinforced memories fade—but never disappear.
When a new memory contradicts an existing one (0.94–0.985 cosine similarity), the old memory is soft-deleted automatically. Your agent's knowledge stays consistent.
1024-dimensional embeddings via Voyage AI. Semantic search that understands meaning, not just keywords. Powered by pgvector for production-grade performance.
No SDKs, no GraphQL, no WebSockets. Just HTTP with JSON. If you can write a curl command, you can integrate Smara in under five minutes.
We're onboarding developers now. Drop your email and we'll send you an API key.
No spam. Just your API key when it's ready.