Simple, transparent pricing.
Start with a 14-day pilot. Scale to Cloud or Enterprise. Agent usage — input and output — is metered separately so you only pay for what you use.
Pilot
Credited to Enterprise contract
A 14-day, fixed-scope production pilot in one workflow — claims, policy, RFP, or compliance. Measured results against your own baseline.
Start a PilotWhat's included
- 2–3 source system connectors
- Up to 50,000 documents indexed
- 1 production workspace
- API access included
- Conflict detection report
- Baseline + results metrics report
- Standard support
- Pilot SLA
Cloud
per month, billed annually
Production deployment on Keel-managed cloud. Full trust layer — source authority, conflict detection, access control, and audit trail — in hours, not weeks.
Talk to SalesWhat's included
- Up to 10 source connectors
- Up to 500,000 documents indexed
- Unlimited workspaces
- Full API access + streaming
- Conflict detection + resolution
- Immutable audit log
- SSO / SAML
- Permission inheritance
- 99.9% uptime SLA
- Priority support (24/5)
Enterprise
Private cloud or on-premise
Full deployment across your knowledge infrastructure. Private cloud VPC or fully air-gapped on-premise. Pricing built from named variables — no per-seat fees.
Contact SalesWhat's included
- Unlimited source connectors
- Unlimited documents indexed
- Unlimited workspaces
- Private cloud (VPC) or on-premise
- SCIM provisioning
- CMEK / bring your own keys
- Custom SLA (up to 24/7)
- Named support engineer
- Quarterly business reviews
- Custom data residency
Input & Output agent pricing.
Every plan includes a usage allocation. Beyond that, input agents (ingestion) and output agents (retrieval) are metered separately — so you scale exactly what you use.
Input Agent
Ingestion & IndexingKnowledge flowing into Keel
Charged when content is ingested, parsed, chunked, and indexed from your connected source systems. Covers entity extraction, relationship mapping, and graph construction.
Re-ingestion on change counts toward monthly volume. Unchanged documents are not re-billed.
Output Agent
Retrieval & ServingContext flowing out to AI
Charged per context query served to AI consumers — copilots, agents, search interfaces, and API calls. Each query returns a structured, cited, conflict-resolved response.
Streaming responses count as one query. Cached responses within 60s are not re-billed.
Want a volume estimate before committing?
Share your approximate document count, query volume, and number of connectors — we'll give you a rough monthly bracket.
Everything side by side.
| Feature | Pilot Fixed fee | Cloud From $2,500/mo | Enterprise Custom |
|---|---|---|---|
| Source connectors | 2–3 | Up to 10 | Unlimited |
| Documents indexed | Up to 50K | Up to 500K | Unlimited |
| Workspaces | 1 | Unlimited | Unlimited |
| API access | |||
| Streaming API | |||
| Conflict detection | Report | ||
| Immutable audit log | |||
| Permission inheritance | |||
| SSO / SAML | |||
| SCIM provisioning | |||
| Private cloud / on-prem | |||
| CMEK / BYO keys | |||
| Custom data residency | |||
| SLA | Pilot SLA | 99.9% | Custom |
| Support | Standard | Priority 24/5 | 24/7 named |
Common questions.
Is pricing based on number of users?
No. Keel pricing is based on knowledge volume, API query volume, connectors, and deployment model — not headcount. Adding users to a workspace does not increase your bill.
How do input and output agent charges work?
Input agent usage (ingestion) is charged per 1,000 documents processed when content is indexed or re-indexed on change. Output agent usage (retrieval) is charged per context query served to your AI consumers. Both have generous included tiers within your plan.
Is the Pilot fee refunded if we move to Cloud or Enterprise?
The pilot fee is credited in full against your first-year Cloud or Enterprise contract — applied at signing, not as a refund.
Can we run on our own cloud infrastructure?
Yes. The Enterprise plan supports private cloud VPC deployment (AWS, Azure, GCP via Terraform) and fully air-gapped on-premise via Helm/Kubernetes. Your data never leaves your cloud boundary.
What counts as a "query" for output agent billing?
Each call to the /search, /context, or /chat endpoint is one query. Streaming responses count as one query regardless of response length. Identical queries returned from a 60-second cache window are not rebilled.
How long does Enterprise deployment take?
Cloud-hosted Enterprise: 3–6 weeks from contract signing. Private cloud VPC: 6–8 weeks. On-premise: 8–12 weeks. Timelines depend on IT access and source system complexity.
Get Started
Stop letting your
AI guess.
See how Keel gives your AI agents, copilots, and applications the verified context they need to answer correctly — every time, with a source you can point to.