Getting started

Quickstart

Create an agent, write something to memory, and ask a question about it. Three requests, about five minutes.

Before you start

You need an account and an API key. Accounts are per region: a key issued in Europe works only against the European host. See Regions if you are not sure which one you are on.

1. Create an agent

An agent is the container that owns memory. Everything you write is scoped to one agent, and one agent never sees another agent's facts.

curl -X POST https://eu.recalld.ai/v1/agents \
  -H "Authorization: Bearer $RECALLD_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"name": "support-bot"}'

The response carries the agent's id. Keep it: every memory call below is nested under it.

2. Write to memory

Send raw content, not facts. Recalld extracts facts and compares them with existing memory. New information can update or supersede older facts.

curl -X POST https://eu.recalld.ai/v1/agents/$AGENT_ID/memory \
  -H "Authorization: Bearer $RECALLD_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "inputs": [
      {
        "kind": "USER",
        "author": "Alice",
        "content": "Bob prefers morning meetings and uses a standing desk.",
        "content_type": "text/plain"
      }
    ]
  }'

The response confirms the write completed. It does not return the extracted facts. Read them back with a retrieval call or from the Threads page.

3. Ask a question

recall is the full pipeline: decompose the question, search, rerank, then select only the facts that answer it.

curl -X POST https://eu.recalld.ai/v1/agents/$AGENT_ID/memory/recall \
  -H "Authorization: Bearer $RECALLD_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"query": "What desk setup does Bob use?"}'

Where next

• Ingestion covers input kinds, authorship, and dating content correctly.
• Retrieval explains when to use recall and when to use search.
• API reference is the complete endpoint list.