Letting people use AI
You do not want to ban every new application. Preparation, review and correspondence are exactly the work where AI wins time, and your people notice that themselves.
AI governance for accounting firms
Your firm already uses AI. Can you account for how?
See which AI tools are actually in use and decide per tool what fits your firm. Help people while they are handling client data, and then show that the policy exists outside the policy document too.
14 days free, no credit card. Rolled out across your organisation within the hour.
"Awareness fades when you're busy. That's why we use BeeSensible." Debby Elfrink, Operations director, Speck Pompen
Your firm 
ChatGPT Limited use 
Microsoft 365 Copilot Allowed 
Claude 
Perplexity No decision yet
ChatGPT Limited use Microsoft 365 Copilot Allowed Claude Perplexity No decision yet Twinfield Allowed
Silverfin
Basecone
Exact Assistant
Boekie AI
Client name Payroll record Financial statements Bank account number Tax file
Your firm Microsoft 365 Copilot Allowed Twinfield Allowed Silverfin Basecone
Client name
Payroll record
Financial statements
Bank account number
Tax file
Plus everything else that is running
Microsoft 365 Copilot Allowed Twinfield Allowed Silverfin Basecone No prompts stored. No per-employee view. Everything processed inside the EU.
What this is about
Confidentiality is part of the profession. AI is now part of the working day.
Someone has ChatGPT rewrite a client email. A tax adviser asks Claude to summarise a file. Somebody tries a new tool for contract review. None of this is unusual any more; these are small efficiencies inside an ordinary day.
For the firm that raises a different question. Not whether AI is used, but which AI is used, what information ends up in it, and whether the agreements you made are holding up in practice.
Protecting client trust
Financial statements, tax files, payroll, contracts and correspondence hold information clients entrusted to you, and not to a vendor sitting behind you.
Being able to account for it
A policy matters, but it does not tell you what happened on a Tuesday afternoon. When the question is what you did to safeguard this, what actually happened is what counts.
Policy and practice
The policy is on the intranet. The decision happens in an input field.
Firms have rules, training and approved applications. What is usually missing is the connection: sight of the tools that quietly become part of the work, and someone beside the person when a client file goes into an AI tool.
BeeSensible makes that connection, in four steps that keep coming back around.
01 See 02 Decide 03 Help 04 Substantiate
01See
Know what your firm has picked up along the way.
A new AI tool does not need to pass through IT to become part of the work. BeeSensible sees when AI tools are opened and shows, at firm level, which ones are genuinely in use, without reading prompts and without building a file on anyone.
- Which tools are in use, and by how many people?
- Which ones appeared for the first time this month?
- What share of the usage runs through what you approved?
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
Daily3428%Weekly3025%Occasionally1714%Not active3933%
Computed on-device and reported as anonymous counts, never per person.
What AI is used for
Analysis & knowledge354
Summarizing353
Writing communication258
Detections over timeLast 30 days
12,438+18% vs the previous period
Detections by platform
78% adjusted
8,124
81% adjusted
3,210
74% adjusted
812
Most encountered data types
4,212
2,930
1,486
1,104
612
SensitiveHighly sensitive
An impression of the AI tools catalogue. The tools and their risk scores come from the product's own catalogue; the verdict in the last column is your firm's to set.
The catalog
865 AI tools assessed for risk, and more arrive every week
You should not have to research a tool before you can decide about it.
Before you can decide anything about a tool, you need to know what it does with your data: where the vendor sits, whether it trains on what you type, whether incidents are known. We worked that out for 865 tools, and more arrive every week. What is left is the question only you can answer: is this allowed here, and under what condition?
The GDPR asks for a record of processing, NIS2 for a view of your suppliers, and the EU AI Act for demonstrable AI literacy. Because every observed tool is recorded with its vendor, hosting region and training policy, and carries a decision from your organisation, you can answer at any moment.
See the catalog-
Which tools are in use, and for what
Every observed tool is in the register, with vendor, country, hosting region, training policy and the kind of work it is used for.
-
Which of them fit your policy
One decision per tool: allowed, limited use, not allowed, or no decision yet. Nothing is silently approved; the decision is always yours.
-
What you do to uphold that policy
Anyone opening a tool you have decided something about sees it at that moment, right in the page. That the notice appeared is counted, without content and without a name, so you can show the mechanism actually reached people.
The overview does not establish that you comply with the law. It shows which measures you can substantiate with measurements, and which you cannot. The judgement stays with you.
02Decide
Not every AI tool deserves the same answer.
A tool that summarises public professional literature calls for a different judgement than one where people process client files. Per tool you see the vendor, the country, the data processing agreement, certifications, documented incidents, a risk score across six dimensions and a separate reading under the EU AI Act. We supply the facts; the firm records the decision, in one of four states.
- Allowed
- This fits inside our boundaries.
- Limited use
- Use it, but not with client data.
- No decision yet
- On the list, not yet assessed.
- Not allowed
- With an alternative that does fit, where one exists.
03Help More about document anonymisation
Bring the decision to where the work happens.
When someone opens a tool you have decided about, they see that decision with the reason behind it, and the alternative that does fit. When they paste client information into a supported AI tool, the sensitive details in the text get marked. Masking or removing happens before they carry on.
BeeSensible does not block and changes nothing by itself. The person stays the one who decides, which is exactly why it works: help you are allowed to ignore does not get routed around.
BeeSensible notice
No decision yet No decision yet on Boekie AI
Your organisation has not made a decision on this tool yet. Do not share personal data or confidential business information here.
Use instead
Open SilverfinAlso allowed: Twinfield ·Basecone
Continue anyway
Only enter public information here
BeeSensible notice
Not allowed Boekie AI is not allowed here
Your organisation does not allow this tool for work. Use an allowed alternative.
Use instead
Open SilverfinAlso allowed: Twinfield ·Basecone
Hold to continue anyway
BeeSensible notice
Limited use Boekie AI is allowed here, with limits
Your organisation allows this tool with limits. Only use it with anonymised data.
Use instead
Open SilverfinAlso allowed: Twinfield ·Basecone
Hold to continue anyway
GeminiDefault1 / 3Draw box
Payroll journal - second quarter
Employee Ilse Draaisma, staff number PN-40917, national ID 139274618, employed by Veldkamp Groep.
Gross monthly salary 4,180 euro, payable to NL91 ABNA 0417 1643 00. Reconciliation with the journal entry has been checked.
7 items will be removedPreviewAnonymise (7)
04Substantiate
See whether your policy exists outside the policy document.
How often was a marking acted on? How often did someone take the approved alternative? What share of AI usage now runs through tools you said yes to? And which kinds of sensitive information still keep turning up?
That movement becomes visible in counts and patterns, without ever showing what an individual wrote. Not to prove compliance, but to back up your measures and steer where it matters.
Warnings about tools
Shown the moment someone opens a tool you have not decided on, or have limited or disallowed. About the tool itself, not about what gets typed into it.
148
warnings shown
61%
chose an alternative
Not allowed52 times
38 alternative9 held through
Limited use44 times
21 alternative15 held through
No decision yet52 times
31 alternative14 clicked through
alternative openedcontinued anyway
Browser use only. Counted at most once per tool per day. Group figures only, never who.
Sent despite warning
1,120
Detected, not adjusted, and still sent.
Critical sent despite warning
214
14% of critical detections
OutcomesLast 30 days
9,702
adjusted before sending
78% of 12,438 detections
Masked7,428(60%)
Removed2,274(18%)
Sent despite warning1,120(9%)
No action recorded1,616(13%)
Accountability GDPRNIS2
- Substantiated
Data minimisation at input
Sensitive data marked and handled before sending
- Substantiated
Register of AI services
Vendor, classification, and a decision per tool
- Partial
Appropriate technical measures
Detection active on the supported applications, handling still below the norm
- Not substantiated
Personal data breach notification
BeeSensible provides the signal, not a reporting process to the regulator
See which AI tools run in your firm within a day.
Roll BeeSensible out on a handful of desks and watch what is already happening. Deciding, helping and substantiating follow after that.
Who looks at this
Three people in the firm, three questions
The partner or the board
Can AI stay, without the firm carrying the risk? You see which tools are running, record a verdict per tool, and keep the time your people are already saving.
Compliance or the DPO
Which measure can you show when someone asks? You get figures on what happened after a marking, at firm level and without a single prompt ever being stored.
IT and operations
What does this cost us? No proxy, and your network stays as it is. The extension rolls out through Intune and the desktop app installs quietly in the background.
Alongside what you already have
What the policy, the training and the filters do not cover
Most firms have something in place already. Here is what each of those measures does, and where it stops.
An AI policy and a training
Both needed, and both work. They are just not there at the moment someone pastes a file, and afterwards they do not tell you what happened. BeeSensible brings the verdict to that moment and counts what came out of it.
Email filters and DLP
Strong on files, email and network traffic. A file that leaves as typed text in a chatbot usually passes them by. Inline cover at the moment of input exists, but depends on the browser, device onboarding and licence tier.
Shutting AI down
Works on paper. In practice the use moves to personal phones and personal accounts, where you see nothing at all. Which is why BeeSensible marks rather than blocks.
A Wednesday at the firm
Risk rarely looks like risk.
09:14
Compilation practice
Someone asks AI to tighten a note to the financial statements. The client name and the figures are still in the pasted text.
11:37
Payroll
A payroll question goes into an AI tool with its context. Name, salary, national ID and account number travel along, because they happened to be in the email below.
14:08
Tax practice
A tax adviser tries a research tool a colleague mentioned yesterday. Nobody has assessed the vendor, and there is no processing agreement.
16:42
Audit
An assistant has AI summarise notes and brings along more file context than the question needs. It saves half an hour.
None of these people is trying to get around the policy. They are trying to finish their work. That is why help during the work beats one more rule.
The rules you already sit inside
Fits obligations that already apply.
Professional rules
Confidentiality is a fundamental principle
In the Dutch VGBA, confidentiality sits alongside professionalism, integrity, objectivity and competence. Article 19 goes further: the accountant takes a reasonable measure so that people working under their responsibility also observe the confidentiality obligations. Most professional bodies carry an equivalent duty.
BeeSensible can be such a measure, and makes its working visible. It does not replace professional judgement.
GDPR
Personal data gets in through people
Once personal data is processed with generative AI, the GDPR can apply. The Dutch data protection authority points out in its view on generative AI that users enter that data themselves while using it.
That is where BeeSensible sits: making data visible before it is shared, and aggregate insight into what happened afterwards.
Professional guidance
From control to application
The Dutch professional body NBA published AI in Control in 2024, a control model for AI systems an organisation is responsible for. In June 2026 NBA Accounttech and NOREA followed with Guideline 2: AI Applied, about putting AI to work in practice.
Once a firm chooses an AI environment of its own, the control work starts: is it used as intended, and what is used alongside it? BeeSensible makes that part workable. It does not make you compliant with a guideline.
The question that always comes
Guard confidentiality without reading along.
“So to check what my people do with client data, you are going to read their prompts?”
No.
Governance should not create a second confidentiality problem. That is why the product is built so the content never settles anywhere.
The full story in the Trust Center- Detection runs on our own servers at European providers, in working memory, and the text is discarded straight after.
- Choose the cloud and detection runs on our own servers in the EU, in working memory, and the text is discarded straight after.
- Prompts do not reach the dashboard. Not shortened, not as an example.
- The usage data holds no user id. There is nowhere it could be written.
- Reporting is counts and patterns, never a per-employee view.
How it is put together
One process, four parts.
The first three can be bought on their own, so you can start with one.
Before you look further
The questions people ask first.
Does BeeSensible read our client data?
No. The text goes to our own servers in the EU, is checked in working memory and discarded straight after. Neither route involves an external AI service, and neither route stores the text.
Can administrators see which person got a marking?
No. For markings, documents and notices, who it was is not recorded, so it cannot be looked up afterwards either. The dashboard has no per-employee view of any kind. At most the department is stored, and only if you switch those insights on, with a minimum group size.
Does BeeSensible block AI tools?
Never. It also never changes what someone wrote by itself. Your colleague sees what was found and what your organisation decided about that tool, and chooses. If something was sent anyway despite a marking, that gets counted, without content and without a name.
Does BeeSensible decide which AI tools we may use?
No. We supply the facts per tool: vendor, country, processing agreement, certifications, documented incidents, a risk score across six dimensions and an EU AI Act reading. The verdict is yours, in one of four states: allowed, limited use, no decision yet, not allowed.
Can we record our own condition per tool?
Yes. A verdict carries the note your colleague sees the moment they open it. So a tool that is fine but not with client data says exactly that, instead of pointing at the intranet.
Where does BeeSensible run?
On our own servers at European providers: the application and the database at Scaleway in Amsterdam, the models at Hetzner in Germany. European companies rather than a European region of an American cloud, so there is no US parent that can be ordered to hand data over. The text is checked there in working memory and discarded straight after. What reaches the dashboard are counts and attributes, never content.
How much work is the rollout?
Not much. No proxy, no change to your network. The extension runs in Chrome and Edge and deploys centrally through Intune; the desktop app installs silently in the background. With a managed rollout your organisation is up inside an hour.
Does this make us compliant with the GDPR or our professional rules?
No, and nobody should promise that. BeeSensible is a technical measure you can put in place and whose working you can show. The lawful basis, the DPIA, the judgement about what is appropriate in your practice and the professional assessment stay with the firm.
How does this sit next to our AI policy and awareness training?
The policy stays where it belongs, and so does the training. BeeSensible brings the decision to the moment someone is working, and then measures whether it was followed. A yearly training session and a notice in the moment do different jobs.
And when a new AI tool turns up tomorrow?
You see it appear in the overview with the facts attached, and you take a decision about it. That is exactly why the cycle comes back around: it is never finished once.
Rather read first
Working safely with AI, for accountancy firms
A short guide to what passes through AI in the compilation practice, the tax practice and payroll, which agreements hold up in daily work, and what you record for your file and the GDPR.
Rens sends it to you himself. If you would rather see it working, just start the trial.
Start at the beginning
Find out what happens between your AI policy and the desk.
In the trial you see which AI tools are in use in your own firm. From there you can go on to deciding, helping and substantiating, at the pace that suits the firm. From an administration office to a firm with its own tax or audit practice.
14 days free, no credit card. Rolled out across your organisation within the hour.
