Capability-gated, not usage-gated
We never charge per memory stored. That penalizes the accumulation behavior that creates Pensiv's compounding advantage.
- MCP server
- Self-hosted
- Single vault
- PostgreSQL + pgvector
- BM25 + vector retrieval
- Surprise gate (SKIP/FAST/DEEP)
- FSRS-6 importance decay
- 10,000 memories
- Your own model keys
- 1 user
- Everything in Community, plus:
- Multi-vault (3 vaults included)
- Full dual-process retrieval (S1 + S2)
- Analogical retrieval (SME)
- Schema induction + families
- Wiki article synthesis
- Tier 1 + 2 connectors (Slack, Calendar, Email, Jira, GitHub, Confluence)
- BYOK + 1,000 included queries/mo
- Audit logs
- SSO
- 5 users
- Managed or self-hosted
- Everything in Professional, plus:
- On-prem / air-gapped deployment
- Federated vaults
- Tier 3 connectors (Salesforce, HubSpot, ServiceNow, SAP, Workday)
- Decision-Grade Artifact API
- Unlimited users & vaults
- Role-based access control
- Per-user audit provenance
- Cross-vault federation
- Dedicated engineer
- Indemnification
- BAA / DPA
- SOC 2, HIPAA, FedRAMP
Fair-use notes
You'll pay for the AI model separately (OpenAI, Anthropic, or free local models). Pensiv doesn't mark up usage.
We ship carefully, not quickly. Each capability clears real benchmarks before going live.
One database to back up, not five. Pensiv runs on PostgreSQL — if your IT team can back up a database, they can back up Pensiv.
If we ever shut down, the source code is held in escrow and gets released. You'll never wake up to find your memory is hostage.
Pricing FAQ
Why don't you charge per memory?
Because that penalizes the behavior we want to encourage. The 100th memory is exponentially more valuable than the first because it has 99 memories to connect to. Charging per memory punishes accumulation.
We gate by capability: Community gets the basics. Professional gets the intelligence layer (analogical retrieval, schema induction, wiki synthesis). Enterprise gets the compliance layer (on-prem, RBAC, federated vaults).
What counts as a "query"?
An LLM-enhanced query — anything that uses analogical retrieval, HyDE, reranking, or Decision-Grade Artifact synthesis. Basic BM25+vector searches don't count.
Professional includes 1,000 LLM-enhanced queries per month. That's ~33/day. Most teams use 200-500/mo. If you exceed, we bill $0.10 per additional query (roughly our cost at Haiku rates).
Can I use my own LLM keys?
Yes. Professional and Enterprise both support BYOK (bring your own keys). Connect your OpenAI or Anthropic account, or run local models via Ollama. You control the LLM; Pensiv provides the memory layer.
What's the difference between "managed" and "self-hosted"?
Managed: We run the Postgres instance, handle backups, monitor performance. You connect via MCP or HTTP API. Best for teams that want to focus on using Pensiv, not operating it.
Self-hosted: You run Postgres on your infrastructure. We provide the schema, the MCP server, and the cognitive pipeline code. Full control. Best for regulated industries or air-gapped environments.
How does Enterprise pricing work?
Enterprise is custom-priced based on:
- • User count
- • Vault count
- • Deployment model (managed vs on-prem vs air-gapped)
- • Tier 3 connector count
- • Compliance requirements (SOC 2, HIPAA, FedRAMP, etc.)
Defense, financial services, and legal typically start at $100K-300K/year. Clinical and competitive intelligence start at $50K-150K/year.
Do you offer academic or non-profit pricing?
Yes. Academic researchers and non-profits get Professional for free if they agree to:
- • Share anonymized benchmarks
- • Cite Pensiv in publications
- • Allow us to reference your work (no data sharing, just attribution)
Contact us with your .edu or .org email.
What if I need more than 5 users on Professional?
Additional users are $15/mo each (billed annually). Or upgrade to Enterprise for unlimited users.
Can I switch between managed and self-hosted?
Yes. Your data is always in Postgres. Export a pg_dump, restore it wherever you want. Pensiv doesn't lock you in.
Enterprise inquiry
Defense, financial services, legal, clinical — we speak your language. On-prem, air-gapped, federated vaults, full audit trails for ITAR, HIPAA, SEC 17a-4, ICD 203.