Find your RAG leakage paths, then close them in two weeks

The Gateco pilot starts with a RAG access risk review: we run real queries as real users and show you exactly where your AI retrieves data it should not. Then we close the gaps and take you to L3 governance. Not a sales trial. A real deployment.

What the pilot includes

Four concrete deliverables, delivered over two weeks with dedicated engineering support.

1

RAG access risk review

We start by finding where your retrieval can leak. Using the Access Simulator and Live Preview, we run real queries as real principals against your connector and show, side by side, what is returned and what is blocked. You leave with a map of the leakage paths in your current setup, a review of your vector database and identity provider, and a clear picture of what closes the gaps.

2

Guided deployment

Step-by-step implementation from connector setup through IDP sync, policy authoring, and first secured retrieval. We work alongside your engineers, not just hand over documentation.

3

Policy authoring support

We help author your first RBAC or ABAC policies, validate them with the Access Simulator, and produce a policy model document your security team can sign off on.

4

L3 certification

By the end of week two, your deployment reaches L3 readiness: resource-level policy bindings active, identity sync confirmed, and a verified first secured retrieval in your audit log.

Timeline

Two weeks of structured work, followed by a documented path forward.

Week 1

  • Connect your vector database (pgvector, Pinecone, Neon, or your existing connector)
  • Run a RAG access risk review: use Live Preview and the Access Simulator to see what retrieval returns today, and where it leaks
  • Connect your identity provider and complete the first principal sync
  • Run retroactive registration and confirm connector readiness reaches L2

Week 2

  • Author RBAC or ABAC policies covering your primary access patterns
  • Validate policies using the Access Simulator before activating
  • Execute the first secured retrieval and review the audit log
  • Produce security review documentation for your compliance or security team

After the pilot

  • Continue on a Team or Growth plan with no additional implementation fee
  • Documented path to L4 chunk-level policy metadata if your use case requires it
  • Direct support contact for 30 days post-pilot

Who it is for

The pilot is designed for engineering teams building internal AI copilots, search interfaces, or agent workflows where the data being retrieved is sensitive: personnel records, financial data, legal documents, customer information. It is the right fit when your security or compliance team needs evidence of access control before the system goes live.

Common starting stacks: Postgres and pgvector or Pinecone for vectors; Okta, Microsoft Entra ID, or AWS IAM Identity Center for identity. We support all four and can work with other configurations.

What it costs

The pilot program is included with a Team or Growth plan subscription. There is no separate implementation fee for qualified teams.

We work with a limited number of pilot teams each quarter to ensure the quality of support stays high. Applications are reviewed on a rolling basis. If we cannot take your team in the current cohort, we will let you know the expected wait.

Included with Team or Growth

No implementation fee. Dedicated engineering support for two weeks.

Apply for a pilot slot

Tell us what you are building and which stack you are on. We review applications and respond within 2 business days. Pilot slots are limited to ensure quality of support.

Max 500 characters

Questions before applying? Contact: pilots@gateco.ai