Platform Platform
System
ConceptsEnginePolicy as codeDeclarationsSafe changeGatewaysIntegrationsObservabilityAdministrationSecurityHuman reviewAudit and evidenceData retentionSecrets and data classification
Controls
Registries and documentationAuthentication and authorizationInjection detectionData redactionCode fingerprintingRole and judge checksContent classificationSpend and loop limitsBusiness rules
Solutions Solutions
By what you do
Sell into the enterpriseControl the AI you run
By industry
Financial servicesDigital assetsInsuranceHealthcareLegalUser-generated content
By discipline
AI governanceTrust and safetyRisk and compliance
Cases Cases Embedded control planeSource-code leakTrading agents over MCPLive firehoseRefund assistant
Compare Compare LiteLLMNVIDIA NeMo GuardrailsOPAROOSTAgent Governance Toolkit
Resources Resources
Guides
Enterprise review questionsPrompt injectionAgent and control layerAgent architecturesDecision system mapAI control maturity model
Standards
Standards OWASP Agent Control StandardEU AI ActPMI AI standardNIST AI RMFERC-8004
Book a demo
Platform · Controls

Any scanner catches a key. Almost none catches your code.

A coding agent sends source to a model on every prompt, and it chooses what to send. Writing the policy takes a minute: proprietary source does not go to an external model. Knowing that this block of text is yours is the hard part.

Rules plus an index built from your repositories.

Why a normal DLP tool cannot do it

Sent by the agentSecret scannerFingerprinting
An API keycatches it-
A public library's sourceignores itignores it, correctly
Your pricing engine, verbatimignores itcatches it
Your pricing engine, renamed and reformattedignores itcatches it

The last row is the whole problem: what the agent sends is never quite what is committed, so string matching against a repository finds nothing.

Where each level of secrecy may go

YAML, versioned, backtestable
signals:
  code_fp:
    udf: dlp/code_fingerprint

rules:
  secrecy_top_requires_level_100:
    condition: >
      signals.code_fp.by_category["secrecy-top"].probability >= 0.5
    effects:
      response:
        required_risk_level: 100

  secrecy_medium_requires_level_50:
    condition: >
      signals.code_fp.by_category["secrecy-medium"].probability >= 0.5
    effects:
      response:
        required_risk_level: 50

A rule states a requirement instead of refusing. Top-secrecy code may only reach a provider you run yourself. The middle tier may go to a vendor you hold a contract with. Anything the table does not name goes wherever the endpoint would have sent it. The gateway takes the strictest level required by the matched rules and by the calling agent, then routes to a provider that meets it, or refuses when none does.

What it protects, in practice

The narrow set of organizations where source code is the company: quantitative trading, chip design, defense and aerospace, robotics, game engines. For everyone else this is a nice-to-have. For them it is the reason the coding agent is banned outright today, which costs the business more than governing it would.

Kept separate from redaction on purpose

Personal data and secrets are matched. Proprietary code is recognized. See data redaction.

Related: source-code leak at a large enterprise
Book a demo