A policy document on the intranet
Describes in general terms what is and is not allowed.
Read at onboarding and never again. A tool that is not in it falls under nothing.
AI policy management software
AI policy management software that applies at the moment of work, not only on the intranet.
AI policy management software is often taken to mean a document management system: write the policy, get it approved and signed, keep the versions. BeeSensible does not do that part. It covers the policy as it exists in the work: a decision per AI tool, a condition the organisation writes itself when a tool is allowed with limits, and that decision in view at the moment someone opens the tool. Afterwards you see in figures whether it is followed, without seeing by whom.
app.beesensible.eu/shadow-ai/inbox
2 newly discovered AI tools await your decision
The problem
The policy is approved. Now it has to happen.
Writing an AI policy is an afternoon's work. Making it apply on the day someone with a deadline opens a new tool is the real work. That is where it usually stalls.
The document does not answer the question being asked
The policy says: only use approved tools for company data. The question at the desk is: may I use this one, now, for this? That answer is recorded nowhere, per tool.
New tools fall outside the policy
AI services keep appearing that did not exist when the policy was written. As long as nobody decides about them, nothing applies, and nobody knows that either.
Enforcement becomes blocking or nothing
Anyone who wants the policy to apply chooses between closing the proxy and hoping. One moves the use to the phone, the other leaves the policy on paper.
Features
What to expect from AI policy management software.
The category runs from document management to enforcement. These are the features that belong to a working AI policy, with an honest note on what BeeSensible does fully, partly and not at all.
| Feature | What it does | BeeSensible |
|---|---|---|
| Decision per tool | Record per AI tool what the organisation has decided: allowed, limited use, not allowed or no decision yet. | Yes Everything starts at "no decision yet". Nothing is pre-selected; the risk score from the catalogue is advice. |
| Your own condition | For "limited use", write down yourself under which condition the tool may be used, in the organisation's own words. | Yes One short sentence, shown in the notice when someone opens the tool. |
| Notice at the moment of work | Show the decision in the browser when someone opens the tool, not in a handbook. | Yes Calm for a tool without a decision, with deliberate friction for a tool that is not allowed. Allowed tools show no notice. |
| Alternative first | For a tool that is not allowed, show the allowed alternative as the first button. | Yes The administrator pins a preferred alternative per tool. Continuing is always possible, after a press-and-hold of a second and a half. |
| Decide on several tools at once | Give a series of tools a status in one go. | Yes In the inbox of newly discovered tools and in the catalogue, with filters and selection. |
| Notice on or off per status | Choose for which statuses a notice appears. | Yes One setting per organisation, switchable per status. The notice for "no decision yet" is off by default, the other two are on. |
| Version control of policy documents | Write the policy document, keep versions and publish it. | No BeeSensible does not manage policy documents. The document stays where it is; this is where the decisions that follow from it live. |
| Workflow with approvals and signatures | Have the policy approved by a steering group and signed by employees. | No Other systems do that. A decision per tool is set by the administrator directly in the dashboard. |
| See whether the decision is followed | Per decision, how often a notice was shown, how often the alternative was opened and how often someone continued anyway. | Yes Group figures only, with no user id, at most once per tool per day. Plus the share of AI use through allowed tools, with its trend. |
What it does not do
No document management, no enforcement.
Two things people expect from this category that BeeSensible deliberately does not do. Plus what goes with that.
- No policy document. You write the policy yourself, in Word or on the intranet. BeeSensible records what follows from it per tool.
- No signatures and no approval workflow.
- No blocking. A tool with status "not allowed" stays reachable. The employee sees the decision, sees the alternative, and chooses.
- No automatic decision based on risk. Every status was set by the organisation.
- No figures per employee. How often someone continued counts as a group figure, never with a name attached.
- No notices in the desktop app. Notices appear in the browser; the desktop app measures AI use and shows nothing.
The approach
From policy document to decision to moment of work.
The same four steps, with the policy document as the part the organisation writes itself.
1 See
See which tools ask for a decision
The inbox shows the tools that were newly discovered and have no status yet. The use determines that list, not the policy.
2 Decide
Decide per tool, in your own words
Allowed, not allowed, or limited use with a condition the organisation writes itself. The policy document is the source; the decision per tool is what remains of it in practice.
3 Help
The decision appears where the question arises
Whoever opens the tool sees the decision and the condition, with the alternative first. No handbook, no searching.
4 Substantiate
See whether it lands
How often the notice was shown, how often the alternative was chosen, how often someone continued. As a group, and as a reason to revisit a decision.
and again How the cycle works
Why this approach
Against how AI policy is usually handled today.
Four common ways to make a policy apply. Each has a place; none of the four brings the decision to the moment someone opens a tool.
An acceptable use policy people sign
Records the agreement, with a signature underneath.
Proves that someone saw the document. Says nothing about what has been opened since.
Blocking through the proxy
Closes known AI domains on the network.
Moves the use to the phone and the home network, and does not know today's new tool.
An annual training
Makes people aware on the day of the training.
Knowledge is no guarantee of behaviour. Under time pressure the tool that is already open wins.
Frequently asked questions
What people ask before they choose.
What is AI policy management software?
Software that manages AI policy. In the broad category that often means writing, approving, publishing and signing the policy document. BeeSensible does the operational part: a decision per AI tool, your own condition where needed, that decision in view at the moment of work, and figures on whether it is followed.
Do I need an AI policy first?
No. Many organisations start the other way round: first see which tools are in use, then decide per tool, and the policy document follows from those decisions. If you already have a policy, you translate it into a status per tool.
What does an employee see?
For an allowed tool, nothing. For a tool without a decision, a calm notice asking to enter only public information, with a button to continue, if the organisation switches that notice on. For a tool with limited use, the condition the organisation wrote. For a tool that is not allowed, a firmer notice with the alternative as the first button; continuing takes a press-and-hold of a second and a half.
Can I enforce the policy?
Not with BeeSensible. Continuing is always possible. The figure for how often that happens is the signal to revisit the decision, the alternative or the explanation.
Where does the condition I write go?
With the status "limited use", per tool, one short sentence. That sentence appears in the notice. The statuses "not allowed" and "no decision yet" use fixed wording.
Can I see who does not follow the policy?
No. The figures on notices and use reach the servers without a user id. You see how often a group continued, never who. That is a design choice, also with the works council in mind.
Further reading
From policy on paper to policy in practice.
Turn your AI policy into a decision per tool, and see it appear where people work.
Fourteen days free, every module, no credit card. Start with the tools that are already in use.
14 days free, no credit card. Rolled out across your organisation within the hour.
