PII
Personal information
Names, addresses, social security numbers and other personally identifiable information.
Airia + Claude
Real-time policy enforcement for Claude Enterprise.
Airia checks every prompt and every tool result against your policies before Claude processes it, then returns an allow or deny decision in real time. One organization-level configuration covers Claude chat, Claude Code and Claude Cowork, so a violation is stopped before it reaches Claude.
“AI governance has to operate at the speed and scale of the technology it governs. We were early to support Claude’s Compliance API and moved quickly to bring inference hooks to customers. Now, organizations can enforce their policies inline across Claude Enterprise before a violation occurs. That is what active governance requires.”
What happens before Claude responds.
- Step 01
A prompt or tool result comes in
A user submits a prompt, or Claude receives a result from a connected tool, MCP server, skill or plugin.
- Step 02
Airia checks it against your policies
The inference hook sends the content to Airia, which evaluates it against your organization's policies before Claude processes it.
- Step 03
Allow or deny, in real time
Airia returns a decision instantly. Requests that break policy are blocked before they reach Claude.
Sensitive data, caught automatically.
When someone pastes production credentials into a conversation or uploads a spreadsheet of customer SSNs, you'll know about it.
PHI
Health information
Protected health information subject to HIPAA requirements.
PCI
Payment card data
Card data that must be protected under PCI-DSS.
Secrets
Credentials and keys
API keys, credentials, tokens and configuration data that should never appear in an AI conversation.
From investigation to prevention.
Find what already happened
Activity from Claude, Claude Code and Cowork lands in one monitoring feed, next to your other models and agents. One view, one investigation workflow. Analyze it against your policies.
Stop it before it happens
Inference hooks move enforcement inline, so a policy decision is made before Claude ever sees the request. When something needs a person, Airia triggers an automated runbook.
When a guardrail triggers, Airia responds.
Runbooks handle routine violations automatically and escalate the cases that need human judgment.
- PII detected in conversationthenRoute alert to Slack security channel
- Secrets exposed in uploadthenCreate ServiceNow ticket and notify identity team
- Policy violation patternthenTrigger downstream identity action
- High-severity incidentthenFlag for immediate manual review
Available now for Airia and Claude Enterprise customers.
1,000+pre-configured connectors for the systems and tools your teams already use
- Immediate deployment with no changes to existing Claude workflows
- Configurable guardrails that match your compliance requirements
- Unified AI agent governance across your entire AI ecosystem