Enterprise AI governance and agent security

AI under your control.From prompt to tool execution.

Chrysent is a vendor-neutral, self-hosted control plane for model access, AI spend, sensitive data, prompt and response policy, agent tool calls, and audit evidence.

Request a demoExplore the control plane

Private beta and design-partner previewsFor enterprises and government teamsOperating since January 2026

Conceptual control pathPolicy made visible
  1. 01
    RequestGateway

    Access · budget · attribution

  2. 02
    ContentTouchstone

    Classification · redaction · policy

  3. 03
    ActionBastion

    Tool-call mediation

  4. Evidence

    Audit decisions and outcomes

One connected evidence trail

The Chrysent control plane

Three controls at the boundary. One evidence trail.

Gateway governs model access and spend. Touchstone applies content policy. Bastion evaluates proposed agent tool calls before the harness executes them and never executes tools itself. Chrysent applies one vendor-neutral policy and evidence model across model requests, prompt and response content, and agent tool calls.

  1. 01 / RequestGateway

    Govern model usage.

    Attribute spend by user, team, project, application, or organizational unit. Set budgets and spending alerts across providers.

    Explore AI Gateway
  2. 02 / ContentTouchstone

    Enforce interaction policy.

    Inspect prompts and responses for prompt injection, jailbreak attempts, prohibited content, and sensitive information. Redact sensitive values before they reach a model.

    Explore Touchstone
  3. 03 / ActionBastion

    Bound what agents can do.

    Inspect attempted tool calls before they touch sensitive resources. Apply configurable policies and retain an internal audit trail.

    Explore Bastion
04 / Evidence

Policy decisions, usage, cost, access, transformations, and execution outcomes feed the evidence layer.

Inspect the platform architecture

Start with your evaluation

Start with the system you need to govern.

Bring security, platform engineering, and FinOps into the same evaluation. Start with one problem; add controls as your requirements expand.

Route AAI governance

Control model traffic, spend, data, and audit.

Establish vendor-neutral access, cost attribution, interaction policy, and reviewable evidence across teams and providers.

Evaluate AI governance
Route BAI agent execution security

Control what agents do with real tools.

Apply policy before tool calls reach files, machines, MCP servers, or production systems, and preserve the decision trail.

Evaluate agent security

Security and deployment

Self-hosted deployment, retention, and failure behavior.

Chrysent software runs in customer-owned or customer-controlled infrastructure. Evaluate the controls, support access, retention choices, failure behavior, and current limits directly.

Self-hosted delivery
Chrysent sells only self-hosted software for customer-owned or customer-controlled infrastructure. Retained product data remains there; customer-configured external model providers receive provider-bound request data under the customer's provider agreement. This model supports data sovereignty and residency requirements.
Provider credentials
Production inference uses customer-supplied provider credentials. Chrysent does not sell model inference.
Retention
Choose metadata-only evidence, encrypted retained content, or an explicitly scoped semantic projection.
Failure handling
Unavailable policy evaluation is not silently recorded as “allow”. Customer can select between "strict" or "best effort" policy failure behaviours.
Support access
Support access requires an expiring, scope-limited grant created by your organization and is logged. Chrysent has no standing tenant impersonation path. Your environment your control.

Review the evidence

Read Security & trustReview data handling & deploymentReview benchmark plans and methodology
Beyond the initial control plane

Expand when the foundation is in place.

Extend the same self-hosted control plane with permission-aware knowledge retrieval and policy-controlled sandbox execution. Availability varies by product; each current status is shown below.

Private Beta evaluation

Evaluate Chrysent on one real AI workflow.

Tell us which models, tools, and policies are in scope. We will route the conversation to Gateway, Touchstone, Bastion, or the full platform—whichever fits your evaluation.

  1. You supply one workflow and its models, tools, and policies.

  2. We examine that workflow together in a working session.

  3. Together, we agree the appropriate product and self-hosted deployment path.

Request a demoReview security & data handling