Persistent memory for AI agents.
Governed. Auditable. Yours.
Recall Racoon gives your AI agents memory you can defend. Every claim is validated before it's stored, tracked through its lifecycle, and recalled with full provenance — so your agents stop forgetting, stop contradicting themselves, and pass audit. Run it in the cloud, your VPC, or fully air-gapped.
Capture once.
Recall when it matters.
Recall Racoon gives your AI systems persistent memory. Information enters from multiple sources. Rumi Rummages through it. Your Stash remembers. Your AI recalls.
Deterministic Control Plane
Separate reasoning from memory policy. Safety rules, personality constraints, and authoritative facts are enforced at the control plane, never left to probabilistic drift.
Hybrid Semantic Search
Vector + BM25 + RRF reranking with local ONNX embeddings. Works fully offline. No API keys, no per-token fees, no data leaves your infrastructure.
Claim Lifecycle Governance
Every memory is a lifecycle-managed claim: candidate → confirmed → stale → superseded → conflicted → archived. Full provenance, citation tracking, and human-in-the-loop review.
MCP-Native Integration
Drop into Claude Desktop, Cursor, VS Code, or any MCP client. Thirteen tools exposed over stdio, SSE, or HTTP. Zero config required.
Local-First by Default
SQLite + sqlite-vec for fully offline operation. Optional Qdrant, Postgres, or Cloudflare Workers backends when you need scale. Your data, your rules.
Secret Redaction & ACL
Auto-redact JWTs, AWS keys, SSH keys, and Bearer tokens at ingest. Memory-level access control with composite IDs and group-scoped search spaces.
Rumi handles the hustle.
Your organization is constantly producing new information. Agents discover things. Documents change. Projects evolve. Decisions are made. APIs send updates. People correct earlier assumptions. Rumi works through that incoming stream and helps turn it into organized, reusable memory.
Understand
Rumi examines incoming information and identifies what it is about.
Summarize
Important context can be condensed into useful memory without losing the original source.
Categorize
Rumi helps organize knowledge into meaningful categories and metadata.
Connect
Related memories can be associated so future Recall is not limited to exact words.
Prepare
Information is prepared for semantic retrieval and future use by connected systems.
Organize
The Stash becomes more than stored content. It becomes an organized memory layer.
Rumi works in the background so your people and AI systems do not have to become memory librarians.
Rumi does not replace your AI tools. Rumi helps them remember.
Experience the Stash
with synthetic data
This demo runs entirely in your browser with seeded synthetic data. No account required. No data stored. Explore the Stash, watch Rummage processing, search memories, and see the knowledge graph.
Stash Browser
15 memoriesDeployment requires approval from SRE team via #deploy-requests channel. Runbook: wiki/deploy-v2
Staging environment auto-deploys on merge to main. Production requires manual gate.
Database migrations must run before service deploy. Order: migrate → smoke test → deploy.
Rollback procedure: revert commit, trigger pipeline, verify health checks.
Connection pool maxes at 20 by default. Increase via DATABASE_POOL_SIZE. Monitor: dashboard/db-health
Symptom: 503 errors with "no connections available". Cause: unclosed transactions in v2.3.1.
Fix: upgrade to v2.3.2 (connection leak patched). Verify: pgbouncer active.
Previous fix (v2.2.8) addressed different leak in connection factory. Not applicable.
Right to erasure: submit via privacy@company.com. SLA: 30 days. Verification required.
Data retention: customer data 7 years post-contract. Analytics data anonymized after 2 years.
ADR-0047: CQRS chosen for order domain due to conflicting read/write models.
Trade-off: eventual consistency acceptable (max 2s lag). Not acceptable for inventory—stays CRUD.
Previous attempt (ADR-0021): unified model caused N+1 on reporting. Reverted.
JWT tokens must never be logged. Redaction pattern: /[A-Za-z0-9-_]{20,}\.[A-Za-z0-9-_]{20,}/
AWS access keys detected in memory are auto-redacted at ingest. Alert sent to #security.
Dual-Plane Memory Architecture
The Data Plane stores embeddings and claims. The Control Plane governs admission, conflict resolution, and policy enforcement. Strict separation prevents probabilistic drift.
Agent Layer
Claude, Cursor, Codex, or custom agents via MCP or HTTP
Transport
stdio, SSE, or HTTP. MCP-native protocol compliance.
Control Plane
Policy enforcement, admission control, conflict arbitration
Data Plane
SQLite/Postgres + vector index. Append-only, snapshotted.
Rumi Governance
Multi-probe validators, semantic contradiction detection, 5-tier auto-resolution
Retrieval Engine
Hybrid search, RRF reranking, freshness scoring, confidence thresholds
Lifecycle Manager
Claim state transitions, decay scheduling, supersession chains, audit logging
Transparent, capacity-based pricing
No free production tier. No unlimited promises. Start with the synthetic demo, then upgrade to a plan that matches your scale and compliance needs.
Explore Demo
Interactive synthetic demo. No customer data storage. Evaluate the experience before committing.
- Synthetic demo only
- Stash browser experience
- Rummage log preview
- Recall search preview
- Knowledge graph visualization
- No data persistence
- No production deployment
- No support
Launch
For teams shipping their first production AI agents with governed memory.
- Up to 50,000 memories
- 3 workspaces
- Hybrid semantic search
- MCP stdio + SSE transport
- Rumi governance (3 policies)
- Email support (48h SLA)
- 7-day audit retention
- No SSO / RBAC
Team
For growing teams with multiple agents and collaboration requirements.
- Up to 250,000 memories
- 10 workspaces
- All transports + HTTP API
- Custom Rumi policies
- Team sharing + permissions
- Priority support (24h SLA)
- 30-day audit retention
- Snapshot replication
Business
For organizations with compliance, scale, and custom integration needs.
- Unlimited memories
- Unlimited workspaces
- Custom embedders + rerankers
- SSO (SAML/OIDC) + RBAC
- Full audit log (1 year)
- Dedicated support channel
- 99.9% uptime SLA
- Custom integrations
Private Deployment
Air-gapped, VPC, or on-premise. Full source-available enterprise modules with license-key activation. Your infrastructure, your compliance framework, your data never leaves.
Contact Sales for Private DeploymentAll plans: Memory, processing, and storage limits apply. No unlimited usage. Overage rates available on request. Annual billing saves 15%. Prices in USD.
Production-ready controls
for regulated industries
Deploy Recall Racoon inside your VPC, behind your firewall, with your compliance framework. We don't see your data. We don't train on your data. We can't.
SOC 2 Type II
Audit-ready controls for security, availability, and confidentiality. Target: Q3 2026.
GDPR / CCPA
Data residency controls, right-to-erasure workflows, and DPA on request.
ISO 27001
Information security management certification. Target: Q4 2026.
HIPAA BAA
Business Associate Agreement available for healthcare deployments on Enterprise Private.
Need a custom deployment?
Forward-deployed engineers, custom integrations, dedicated support channels, and bespoke SLA terms. We work with your procurement team, not around them.
How we differ from
general-purpose memory
Other memory systems accumulate. We steward. Here's what that means in practice.
| Capability | Recall Racoon | General AI Memory SaaS | Vector DB + Custom RAG | Knowledge Graph Memory |
|---|---|---|---|---|
| Unit of memory | Claim with lifecycle + citations | Text chunk / summary | Entity / relationship | Document chunk |
| Stale fact handling | Decay → stale → superseded | Overwrite (silent) | Versioned (manual) | Lingers until deleted |
| Conflict detection | 5-tier auto + manual queue | None (add-only) | Graph-level only | None |
| Provenance / citations | Per-claim citation + agent trace | Usually none | Source URL only | Chunk source |
| Secret redaction | JWT/AWS/SSH at ingest | Your problem | Your problem | Your problem |
| Offline / air-gapped | Full operation, zero network | Cloud-dependent | Self-host possible | Depends on stack |
| MCP-native | Built on MCP, not wrapper | SDK / wrapper | SDK / wrapper | Build your own |
| Cost model | Capacity-based, predictable | Per-token / SaaS | Per-token / SaaS | Variable (infra + API) |
Common questions
No. The interactive demo runs entirely client-side with precomputed synthetic data. No LLM inference, no vector writes, no storage, no data leaves your browser. It is a product simulation for evaluation purposes only.
You'll receive quota notifications at 80% and 95% capacity. At 100%, new writes are rejected with a clear error. No silent data loss. Overage is available at a published per-unit rate, or you can upgrade to a higher tier.
Yes. Private Deployment supports fully air-gapped operation with license-key activation and a 30-day grace window. No network calls required after initial setup. Your data never touches our infrastructure.
As information enters Recall Racoon, Rumi Rummages through it—examining meaning, identifying metadata, finding relationships, and preparing it for the Stash. Every memory enters as a candidate. Rumi probes it (format, citation, semantic, contradiction). Valid claims become confirmed. Over time, confidence decays; stale claims are flagged. New evidence can supersede old claims. Conflicts enter a resolution queue. Archived claims remain auditable but are excluded from Recall.
No. Your data never leaves your infrastructure on Private Deployment. On Cloud Managed, we process it but do not retain, aggregate, or train on it. We have no access to your Stash, queries, or embeddings. This is enforced by architecture, not policy.
Launch includes email support with 48-hour response SLA. Team includes priority email with 24-hour SLA. Business and Enterprise include business-hours support with 4-hour response for critical issues, plus a dedicated Slack channel and forward-deployed engineer option.
Sales Inquiries
For pricing, licensing, deployment options, and procurement questions.
sales@recallracoon.comSecurity & Compliance
For security questionnaires, DPAs, BAAs, and audit requests.
security@recallracoon.comWhat to expect
- Response within 1 business day
- Architecture review call offered
- Security whitepaper provided
- Custom quote within 3 days
- Procurement-friendly terms
Private Deployment
Air-gapped, VPC, or on-premise. Full source-available enterprise modules with license-key activation. Your infrastructure, your compliance framework.
Request Private Deployment Info →Tell us about your project
All fields marked with * are required.