Discovering AI Tools

Peridot continuously discovers AI tools used across your organization.

This includes both sanctioned tools and shadow AI usage.

Discovery is automatic and driven by integrations, not manual input.

Why discovery is hard

AI usage is fragmented.

Teams adopt tools independently. AI is embedded into existing SaaS platforms. Internal scripts call model APIs directly. Logs and workflows often don’t explicitly label “AI usage.”

As a result, most organizations underestimate their AI footprint.

How Peridot detects AI usage

Peridot identifies AI tools through multiple signals:

API and model activity

Requests to model providers such as OpenAI, Anthropic, and Google.

Integration behavior

AI usage within connected SaaS platforms and internal systems

Cloud and log data

Signals from cloud logs, observability systems, and event streams.

Application patterns

Workflows and applications that include AI-driven steps.

Types of AI tools discovered

Peridot classifies AI usage across:

External AI tools (e.g., ChatGPT, Claude)
Embedded AI features in SaaS platforms
Internal AI services and workflows
AI-generated applications

Continuous discovery

Discovery is not a one-time scan.

Peridot continuously updates inventory as:

New tools are introduced
Usage patterns change
New integrations are connected

From discovery to control

Discovery feeds directly into:

Inventory classification
Data flow tracking
Policy enforcement

Without accurate discovery, governance cannot function.

What to do next

Review Shadow vs Sanctioned AI to classify usage
Move to Data Flows Overview to track how data moves through these tools