AI OR NOT
Classification
Every new app and domain classified in real time: AI or not, category, kind of AI.
BROWSER · ENDPOINT · IDENTITY · SAAS · CLOUD
PLATFORMDiscover · Classify · Intent · Agents · Control
In the browser, on the laptop, inside your identity provider, SaaS and cloud. Identra finds the AI, understands it, and protects it beyond the prompt, tied to one identity and one timeline.
one identity
§ 01Discover
Find the AI nobody told you about.
One AI inventory across browser, endpoint and cloud: the apps, the accounts behind them, the extensions, the agents and the grants.
AI INVENTORY
BROWSERDISCOVERED
- AI apps and the accounts people use with themWORK · PERSONAL
- Every installed browser extensionEXTENSIONS
- AI agents driving the browserAI AGENT
§ 02Understand
A list of AI isn't enough. Understand it.
What it is, what it's for, how it fits the way you work, and what it can touch.
WHAT IS THIS AI?REAL TIME
FIRST SIGHTINGnew-ai-notes.app
- AI OR NOTAI
- CATEGORYmeeting notes
- KIND OF AIAI app
- CRITICALITYcan reach your data
Classified in real time, as soon as a new app shows up.
§ 03Beyond the prompt
Control what AI does, beyond what it reads.
Policy in the browser and at the tool call. Approval-based response on the grant.
ACCOUNT-AWARE AI ACCESS
WORK
Allow
PERSONAL
Redirect
WHAT THE PERSON SEESALLOWED
ChatGPT with a work account: allowed.
Browser
- Allow, redirect or block AI by signed-in account.
- Prompts allowed, alerted, masked or blocked.
- Uploads to AI blockable.
- Browser-driving agents blockable.
- Risky extensions can be disabled by policy.
Endpoint
- Prompts to Codex and Claude Code blocked when policy is active.
- Agent tool calls checked against policy.
- Destructive shell commands denied when policy is set to block.
- On macOS, access to protected files can be denied.
Identity · SaaS · Cloud
- Risky grants revoked by an analyst.
- Sign-in sessions revoked with approval.
- Every result recorded.
§ 04One timeline
Your AI, traced to a person.
Browser, endpoint and cloud activity on one person's timeline, grouped into incidents. Optional AI triage explains them. It advises. It never acts.
- REVOCATIONS NEED APPROVAL
- ONE NAMED TARGET
- RESULT RECORDED
§ 05Surfaces
Go deeper on each AI surface.
FAQ
Questions, answered.
What does Identra protect?
AI in use across your company: AI apps and accounts in the browser, coding agents and AI clients on macOS and Windows laptops, and connected apps, OAuth grants and AI agents across your identity, SaaS and cloud providers.
Do prompts leave the device for inspection?
No. Prompt checks run on the device before send, and prompt content stays on the device by default.
Which AI apps does account-aware access cover?
ChatGPT, Gemini, Claude, Perplexity and Grok: allow, redirect to your company AI workspace, or block, by signed-in account.
Does Identra's AI take action on its own?
No. Optional AI triage explains incidents using anonymized context. It advises. It never acts. Revoking a grant or a sign-in session needs a person's approval and applies only to the item they chose.
Which providers does Identra connect to?
Microsoft 365 and Entra ID, Google Workspace, Okta, AWS, GitHub, Salesforce, Slack, Anthropic and other providers.
BOOK A WALKTHROUGH
See the AI in your company.
A walkthrough with a security engineer, on your browsers, laptops and cloud.