Developer guide
Long-term memory API for AI agents
Give your agent durable context without rebuilding storage, retrieval, and data controls from scratch.
The basic memory loop
A memory-aware agent writes a useful conversation, waits for processing, then retrieves relevant context before generating its next response.
pip install syntarus
Production controls
- Keep customer data isolated by project
- Use separate read and write scopes
- Wait for event success before depending on new memory
- Use deletion and export controls for user data requests
- Use webhooks when downstream systems need processing updates
When to use a long-term memory API
Use one when your agent serves returning users, handles multi-step work, or maintains context across sessions. For a single-turn chatbot, ordinary prompt context may be enough.
Open the developer portal ? Read the API reference ? Discuss a pilot