Control over AI
Documentation
AI Tools (Shadow AI)

Adoption: how AI is used across your team

See how many people use AI, how often, and for what. Anonymous and at organisation level.

Adoption

Is AI being adopted in your organisation: how many people, how often, and for what.

Refresh
7 days30 days90 days
31 Jul 2026 – 30 Aug 2026
AI adoption
68%
81 of 120 seats AI-active this month
Prompts sent
1,596
22% vs previous 30 days
AI tools in use
7
2 newly discovered this period
Safe adoption
71%
of AI use via allowed tools
How often AI is used
Your team by usage frequency this month. Anonymous counts: the frequency is computed on each device, never per person.
Daily3428%Weekly3025%Occasionally1714%Not active3933%
Computed on-device and reported as anonymous counts. Someone using two browsers may count twice.
Adoption over time
Prompts per week (bars) and distinct AI tools used (line).
0150300450600024681 Jun8 Jun15 Jun22 Jun29 Jun6 Jul
PromptsTools used
What AI is used for
354 prompts · 3 tools
chatgpt.com
168 · 47%
claude.ai
102 · 29%
copilot.microsoft.com
84 · 24%
Adoption shows how much AI your team uses: adoption rate, usage frequency, and what AI is used for this period.

The AI Tools module gives your organisation a clear picture of AI use: is it actually being adopted, which tools carry the risk, and what to do about it. It does not block access: it makes AI use visible, helps you set policy per tool, and guides people towards safer choices.

In the dashboard menu the module sits under AI Governance, with four screens: Adoption (this page, the positive view), Risk (the watchful view), the Inbox for deciding on newly discovered tools, and the Catalog for reviewing risk and setting policy.

What Adoption shows

Four numbers run along the top:

  • AI adoption: how many of your seats were AI-active this month.
  • Prompts sent: every message sent to AI tools in the chosen period, with the trend against the previous period.
  • AI tools in use: the number of distinct AI tools seen in the period.
  • Safe adoption: the share of AI use that runs through allowed tools.

Below that come How often AI is used, Adoption over time (prompts per week as bars, distinct tools used as a line), What AI is used for, and the AI tools used table with each tool's status, visit count, and prompt count. A tool used only through the desktop app has prompts but no browser visits; its visits column shows a dash.

Change the period with the date picker at the top; only the adoption number always looks at the current month. The signal is usage, not text: no message content is stored.

How often AI is used

This card splits your team by usage frequency: Daily, Weekly, Occasionally, and Not active.

The counting is deliberately kept away from individuals. The extension tracks on the device itself on which of the last 30 days AI was used; that list never leaves the machine. Once a month it reports only the outcome, the frequency bucket, as an anonymous count. So the dashboard receives totals per bucket, never who is in one or when anyone did anything.

The breakdown only appears once at least 5 people have been AI-active this month, so a distribution can never describe one person; below that, the card shows only the number of active team members. Someone with the extension in two browsers may count twice, which the card also says in its footnote.

What AI is used for

The use-case panel answers the question behind the counts: not just that people use AI, but what for. Prompts are grouped into categories: Writing communication, Marketing & content, Summarizing, Rewriting & editing, Translating, Coding & technical, and Analysis & knowledge.

What AI is used for
354 prompts · 3 tools
chatgpt.com
168 · 47%
claude.ai
102 · 29%
copilot.microsoft.com
84 · 24%
Pick a use category on the left and see which tools carry it on the right. Anonymous counts, no content.

Expand a category and you see which tools carry that work, with each tool's prompt count and share within the category. That drill-down is what makes the panel actionable: if translating turns out to run through a consumer tool, you know exactly where an allowed alternative would land best.

The category label comes from a one-time classification of the first message of a conversation; the follow-up messages inherit it. When detection runs in the cloud, the classification happens on our EU servers; when detection runs locally, the desktop app does it on the device itself. What gets stored is only the combination of tool, category, and count, never the prompt text. If classification fails, the prompt counts as Unclassified; the totals still count every prompt.

Reading the numbers

Rising adoption with a healthy safe-adoption share means AI is spreading through allowed tools, which is the goal. If use is climbing but safe adoption is not, check Risk for where the gap is. A category climbing in "what AI is used for" tells you where AI is becoming part of how people work, which helps you decide where an allowed, safer tool would land best.

What stays private

The figures on this page are anonymous counts: no message content, no screenshots, and no user id in the AI-usage figures. Adoption reports patterns at organisation level, never individual people; nowhere in the dashboard can you drill down to one employee. The detection side (Realtime Privacy) is completely separate and never passes content to the AI Tools module.