How MIND Integrates With Microsoft Copilot and Purview

Itai Schwartz, Co-Founder & CTO at MIND

Nov 18, 2026

Copilot exposes whatever your permissions and labels allow, so MIND works inside the Microsoft tools you already run.

Most organizations have lived for years with files shared more widely than anyone intended and no sensitivity label to flag them. Copilot changes how fast that problem gets found. It inherits the user's existing permissions, so anything an employee can access, Copilot can surface. Ask "How much does my boss make?" and it might return an HR file that was shared with the whole company by mistake and never got a Microsoft Information Protection (MIP) sensitivity label.

Many security teams can describe their DLP policies. Far fewer can say those policies are enforced consistently, that MIP labels are applied correctly and that sensitive data is protected from exposure through AI.

That's why this partnership matters so much to me. Microsoft is where many of our customers already run their business, and it's where Copilot is landing in their day. Working with Microsoft lets us meet customers inside the tools they've already chosen, instead of asking them to adopt something separate to close the gap. I'm thrilled about what it opens up for the customers we serve.

What the partnership means in practice

MIND and Microsoft have signed a partnership agreement to help Copilot rollouts start with the right data security fundamentals. The integration is the technical work behind it, and it's what I want to cover here. MIND reads the Microsoft signals customers already rely on, including Purview sensitivity labels, Microsoft Graph data and Windows endpoint activity, and adds the context those signals don't carry on their own. A company rolling out Copilot quickly shouldn't have to choose between speed and protecting its sensitive data.

Inspecting Copilot prompts on the device

MIND supports Copilot in the browser and in the native app, and it inspects prompts. On Windows, MIND's endpoint agent applies real-time data loss prevention to content moving into Copilot, so you can roll Copilot out to your people and still stop a sensitive prompt on the device before it leaves. MIND also classifies the files Copilot can reach, so the people running a rollout can see what Copilot would surface before it does. Purview's native Copilot protections stay in place, and MIND adds AI-driven classification and real-time DLP across these Copilot experiences.

Classifying Microsoft 365 content through Graph

MIND connects to Microsoft 365 using the Microsoft Graph API and runs our multi-layer classification engine on all the files in SharePoint, OneDrive and attached to Teams messages. Classification looks at the content and its context, not just pattern matches. The same card number means something different in a finance folder than in a public marketing folder, so context changes the risk. Security teams can start with the files that carry real exposure. The Graph permissions MIND needs are listed in the MIND documentation.

Enforcing policy in Edge and on Windows endpoints

MIND's endpoint agent runs on Windows 10 Build 1607 and later. It applies real-time data loss prevention across Microsoft Edge with no minimum Edge version, so you can stop a sensitive transfer in the browser as it happens, including browser-based Copilot prompts. One rule set covers files, browser activity and Copilot prompts.

The endpoint agent enforces policy on the device at the moment of the action, not in a report the next morning. The goal is to block the risky transfer, explain the decision to the user and leave ordinary work alone. Policies start from defaults and can be tuned to your environment. Log and review each tuning change, because an overly broad rule is the fastest way to lose the trust of the people using the tool.

Identity context from Entra ID

MIND reads Microsoft Entra ID directory information through the Graph API, including groups, users, phone numbers, organizational trees and activities. That adds identity context, so the people who access sensitive data can be used as a point of policy enforcement. A sensitive file shared with one team gets a different response from one open to every employee, and identity is what makes that difference visible.

How MIND reads and writes Purview labels

Purview is the data security platform MIND works with directly, and the goal is to help it work better for your organization. Sensitivity labels record how your organization decided a file should be handled. MIND doesn't replace that decision. Using the Microsoft Information Protection SDK and the Microsoft Graph API, MIND reads the label on any file that carries one, across every file type Microsoft supports.

Reading the label lets MIND compare it with what's actually in the file. A label is a decision made at one point in time, and the file may have changed since. Where your policy calls for it, MIND can also apply labels, and its classification informs the label instead of sitting beside it. Your organization keeps the labels it has invested in and can see whether each one still fits the material it protects.

When our multi-layer classification engine scans a file, your labeling policy can apply the right MIP label automatically, based on the sensitive content it finds. Customers tell us users don't always apply the correct label, and files change after a label is applied. Now supported files get the label their content and context call for, without anyone reminding end users.

Email DLP for Exchange

With Microsoft Exchange, we inspect message bodies and attachments, classify sensitive information in organizational mailboxes and prevent it from leaving the organization through email. Email is a fast path out of an organization, so it gets reviewed in the same pass as files, endpoints and Copilot prompts.

Closing exposure before Copilot finds it

Start with an inventory. Find out where sensitive data lives across Microsoft 365 and endpoints before a Copilot pilot widens access. Then check how well your labels match what the files hold. Discovery and labeling are often the parts that fall behind schedule, and getting both right moves the rest of the program faster.

Context also changes urgency. A finding on a file one department can open is a different problem from a finding on a file anyone in the company can reach, which is why discovery and access data need to be read together, so you can fix the riskiest exposure first. Findings can drive remediation. Permissions can be adjusted, and issues can be escalated to the data owner, who hears about the exposure first. Real-time controls on endpoints and email then keep sensitive data from leaving.

Purview's controls stay in place, and Copilot still respects existing user permissions, so MIND works on top of those boundaries. A simple test helps. Run both, compare what each one flags and look at what neither one flags. Document those gaps, assign an owner to each and revisit the list after your first full discovery cycle. Most of the surprises in a Copilot rollout live in that last group.

Thank you to the team

Major kudos to everyone at MIND who made this integration real, and to our colleagues at Microsoft who worked with us to get here. The Graph, endpoint, email and identity work in this post took careful engineering, and I couldn't be prouder of what the team delivered.

Test it against your own tenant

If you're working through a Copilot rollout, judge this integration against your own tenant. A MIND demo walks through these connections with your data security priorities, so you can see what each layer adds before anything changes in production. MIND is also available in the Microsoft Marketplace.

DLP you have to see to believe

Request a demo