Information security & cyber

ADRI
Let AI work. Keep control.

ADRI is a security and observability platform for AI use in the organization: discover chats, agents and MCP servers, apply policy that allows, redacts or blocks in real time, and keep a record of every decision.

About the product

What is ADRI?

Employees already use ChatGPT, Claude, Gemini and Copilot, and developers work with coding agents such as Claude Code, Cursor and Codex that connect to tools and data through MCP servers. Any prompt can become a leak path for customer data, credentials and company knowledge, and any agent can take actions beyond the conversation.

ADRI works at the point of AI use: a local agent and a browser extension, deployed through your existing device management. It discovers the AI tools in use, checks prompts, files and tool calls against your policy, and keeps work moving - redacting sensitive values or blocking a prohibited action.

ELPC deploys ADRI for customers: from mapping current AI use, through a staged monitor-only rollout, to policy design and integration with the organization's security and compliance tools.

Capabilities

What you get with ADRI.

Shadow AI discovery

Map the AI tools, agents, extensions and MCP servers in use - who uses them, where and in what context.

Sensitive data protection

Inspect prompts and files and redact sensitive values before they reach the model, instead of blocking the work.

Agent & MCP security

Approve MCP servers, define which tools an agent may call and stop unauthorized actions.

Runtime AI policy

Contextual decisions: allow, redact or block by user, tool, data type and action.

AI threat detection

Correlate AI, EDR, SaaS, device and extension signals into one investigation view.

Observability & evidence

Every decision recorded: rule, severity, action, user and device - a basis for audit and AI governance.

How it works

From discovery to protection, one flow for all AI use.

1. Discover

Chats, agents, MCP servers, extensions, IDEs and SDKs land in one AI inventory with user and device context.

2. Govern

Define which tools are approved, how sensitive data is handled and which agent actions are allowed.

3. Protect

Each supported interaction is checked in real time: allow, redact or block, and the work continues.

4. Investigate & record

Every decision is kept with its context, and suspected threats become an investigation-ready story.

Supported tools & rollout
ChatGPTClaudeGeminiMicrosoft CopilotCursorClaude CodeCodexMCPSlackGitHubMicrosoft TeamsNotionAtlassianIntuneJamfKandjiSCCMGPOEDRSIEM

Why existing security tools are not enough for AI

Cloud tools only see sanctioned apps, and network monitoring cannot see inside an encrypted chat or an agent running locally. By the time the data reaches the AI provider, the alert is too late.

ADRI does not replace your EDR, proxy or DLP, and does not require traffic rerouting. It adds an AI-specific layer where the employee or the agent actually works.

  • Employees keep the tools they like, without sensitive data leaving.
  • Coding agents work within clear limits: what they may read, write and run.
  • The security team gets one picture of all AI use instead of guesswork.

Privacy, hosting and regulation

By default evaluation runs in the customer's cloud tenant; optional on-device evaluation keeps supported prompt content on the device and reports only violation metadata. The Israel-hosted SaaS runs on AWS in the Israel region with a separate database per customer. Self-hosted and isolated deployments are also available.

ADRI provides controls and evidence that support an AI governance program and Amendment 13, GDPR, ISO 27001, ISO 42001 and the EU AI Act. Using the tool alone does not make an organization compliant - ELPC also covers the organizational side: policy, procedures, training and risk management.

How to start

  • Discovery call: which AI tools are in use, what data is sensitive and which regulations apply.
  • Focused pilot: deploy to a group of users in monitor mode and see actual usage.
  • Staged policy: redact sensitive data, approve tools and govern agents.
  • Roll out across the organization with ongoing ELPC support.
The ELPC difference

Not just a license - a managed service.

The product is only half the solution. ELPC’s team designs, deploys and operates it as part of your IT and security department.

  • Mapping current AI use and an initial gap report
  • Rollout through Intune, Jamf or GPO, monitor-only first
  • Writing an AI use policy and turning it into platform rules
  • Integration with your EDR, SIEM and ticketing
  • Alignment with Israeli Amendment 13, ISO 27001 / 42001 and the EU AI Act, including vendor due diligence
FAQ
What is ADRI?

ADRI is a security and observability platform for AI use in the organization: discovery of AI tools, agents and MCP servers, policy that allows, redacts or blocks supported interactions in real time, and a record of every decision.

Does ADRI block AI use?

The goal is safe use, not blocking. Often ADRI redacts the sensitive value and leaves the model a usable prompt; blocking is kept for actions the policy prohibits.

Does prompt content leave the device?

It depends on the mode. With cloud evaluation (the default) content is sent to the customer's tenant over an encrypted channel. With on-device evaluation supported content stays on the device and only violation metadata is reported.

How is ADRI deployed?

A local agent and a browser extension are deployed through your existing device management - Intune, Jamf, Kandji, SCCM or GPO - or installed directly. Rollouts usually start in monitor-only mode.

Does ADRI replace our EDR or DLP?

No. ADRI works alongside your existing security tools and adds an AI-specific layer. It can also enrich investigations with EDR, SIEM and ticketing context.

Does ADRI help with Amendment 13 and the EU AI Act?

ADRI provides controls and evidence that support your governance and compliance program, such as redaction of personal data, tool governance and decision records. Compliance also requires a broader organizational assessment, which ELPC supports.

Related products

Product names and trademarks mentioned on this page are the property of their respective owners. ELPC provides consulting, deployment and managed services for these products.

30 minutes with an expert

Ready to turn IT from a ticket queue into a system under control?

30 minutes with an ELPC expert - a clear picture of your infrastructure, gaps and risks. No sales deck.

ISO 27001
ISO 27799
ISO 9001
Approved supplier, Israel Ministry of Defense

We use this information to contact you and respond to your enquiry. Together with your enquiry we also record how you reached the site (for example the page, campaign or network you came from, your device type and approximate country) to measure and improve our marketing. You are not obliged to provide it, but without it we cannot get back to you. You may ask to access, correct or delete your data - see our privacy policy.

Chat with us on WhatsApp