Start with local capture and search. Add cloud sync, shared knowledge, personas, tickets, AgentRuns, and self-hosted deployment as your team matures.
Free for individual local use. Team and enterprise plans scale with storage, governance, and deployment requirements.
These tiers map to how teams adopt agents: local memory first, shared memory next, then governed coordination and enterprise deployment.
Self-hosted Kubernetes deployment on your infrastructure. Your data never leaves your network.
HIPAA-ready, SAML SSO, SLA, dedicated support, and air-gapped deployment options.
Yes. Local capture, resume, and search are free. Paid plans add hosted sync, team-wide knowledge, personas, tickets, AgentRuns, service keys, governance controls, and enterprise deployment options.
Session data is the actual cost driver — not headcount. A solo developer running intensive sessions can generate more storage than a large team of occasional users. Storage pricing is honest: you pay proportional to what you actually use and what it costs us to store and serve.
Only for LLM-powered features like the Judge and narrative summaries. Bring your own key (BYOK) — it is never stored on our servers. Works with any OpenAI-compatible endpoint including OpenRouter, LiteLLM, vLLM, Ollama, and Azure OpenAI. Deterministic features like session capture, resume, search, and basic summaries require no API key at all.
The daemon defaults to local-only — sessions stay on your machine. Cloud sync is an explicit opt-in. You can push individual sessions or enable auto-sync. Enterprise users can self-host the entire backend so nothing ever leaves their network.