Nudger — AI governance

Your staff are already using AI. Make it governable.

Nudger is AI governance: it stops confidential data leaking into AI tools across the browser, the IDE and AI agents. It finds out which assistants and models are genuinely in use, applies the rules you set about what data can go where, catches sensitive data before it leaves, and tells you when practice starts drifting from policy — while people are still working, not at the annual review.

Book a demo View pricing

£5.99/user/month (suggested minimum £19/month) — no EDR required.Upgrade to AIOpenSec Complete at £9.99/endpoint/month when you need the full stack.

AI USE — DISCOVEREDNUDGER

Assistants in active use11

Approved in your policy4

Handling customer data3

Teams with no named owner2

Policy last reviewed14 months ago

Illustrative figures. The first run is usually the uncomfortable one.

HOW IT WORKS

Inventory, rules, then the nudge

01 — INVENTORYSee what is actually in useBrowser extensions, desktop apps, API keys charged to a team card, models embedded in tools you already bought. The list is always longer than the one people give you when you ask.

02 — RULESWrite rules people can followWhich data classes may go to which tools, who owns each approval, what needs review before it ships. Set once, enforced in the places work happens.

03 — NUDGECorrect at the moment, not afterWhen someone is about to paste client data into an unapproved tool, they get a short prompt and a sanctioned alternative. Blocking makes people route around you; a nudge changes the habit.

WHERE IT RUNS

Governance by surface, not a blanket claim

Coverage depends on how a tool is used and how Nudger is deployed. We describe what each surface actually covers, rather than claiming total visibility over every possible AI interaction.

BrowserCoverage of AI tools used through the browser — discovery, policy enforcement and contextual nudges at the point of use.

Thick-client / proxyNudger Proxy extends governance to supported desktop applications and thick-client AI tools, not just the browser.

IDEA VSIX extension brings the same policy and nudging into the editor, for AI assistants used directly in development.

MCP and agentsWhere MCP servers and agent tool-connections are deployed, they are inventoried and governed under the same rules as any other AI integration.

Why we did not build a blocker

The first version of this product blocked unapproved tools outright. Within a fortnight, pilot users were doing the same work on personal laptops where we could not see any of it. Governance that pushes work into the shadows makes an organisation less safe, not more. So Nudger informs, offers the approved path, and records the decision.

FOR RISK AND COMPLIANCEAn evidence trail showing the policy exists, is applied, and where exceptions were granted. Sensitive data is caught with PII detection and redaction before it leaves, not flagged after the fact. Domain-level threat intelligence (MISP) also flags exposure tied to your organisation as a whole, not filtered down to one mailbox.

FOR ENGINEERINGModel and API use per service, so you find out about a new dependency before procurement does — including MCP servers and tool connections, governed the same way as any other AI integration.

FOR LEADERSHIPOne page you can take to a board: what we use, under what rules, and where we are exposed.

Bundled, or on its own

AI governance is one of the six functions inside CyberFit. If that is the only piece you need right now, Nudger is available separately — and it upgrades into the platform later without a migration.

See CyberFit Talk to us about Nudger