Everyone on their own subscription, and the company none the wiser
AI tools for individual use are cheap, good and easy to sign up for. That is why they walk into companies through the back door.
Updated
Two ways to buy AI
An individual subscription is built for one person. Acotia is built for a workforce. Neither is better: they answer different questions.
| Individual multi-model subscription | Acotia | |
|---|---|---|
| Who buys it | Each person, on their own card or the company's | The company, with its own provider accounts |
| Sensitive data | Leaves exactly as the person typed it | Detected and masked, sent to a European model or blocked before it leaves |
| What the company sees | Nothing, other than the card spend | Usage, cost and risk avoided, without reading the chats |
| Spend | Separate fees and credits per person | Caps per person, team and company; consumption goes to the company's account |
| When someone leaves | The account stays theirs, and so does the history | Suspending or removing them cuts access immediately |
| Proof in an audit | No company record | Evidence log without content, exportable as CSV |
If your team is three people and handles no third-party data, an individual subscription is probably enough. Acotia starts to make sense when some of the data is not yours and someone has to answer for it.
What happens without anyone deciding it
Someone tries a tool with models from several providers, it works for them and they tell their team. Within a month there are ten subscriptions, no contract and no policy. Nobody did anything wrong: the company simply never offered an alternative.
Where the problem is
It is not the price of each fee. It is that client documents, contracts and payslips pass through accounts the company does not control, cannot audit and cannot close when someone leaves. For a law firm, an advisory firm or a clinic, that is a confidentiality matter before it is a budget one.
What a company environment does differently
The team keeps using good models and a similar chat interface. What changes is the layer underneath: the data policy applies before text leaves, usage is measured without reading what is written, and spend has a cap. And when someone is removed, access is cut at once.
How to start without banning anything
Banning personal tools rarely works: people keep using them on their own. Offering an equally comfortable alternative works better. That is why the pilot starts in audit mode: it measures how much AI is really used and what data passes, before deciding what to mask or block.
Frequently asked questions
Is it a problem if my employees use personal AI subscriptions?
It depends on what they type into them. If they handle client data, contracts or payslips, that data passes through accounts the company neither controls nor can audit. If they only draft generic text, the risk is low.
Should personal tools be banned?
It rarely works. It is more effective to offer an equally comfortable environment with the data policy applied, and to measure usage before deciding what to restrict.
When is an individual subscription enough?
When the team is very small and handles no third-party data. Beyond that, the lack of a record, spend caps and leaver control usually weighs more than the price of the fee.
Does Acotia give access to several models?
Yes: OpenAI, Anthropic, Google and Mistral (EU), using the company's own accounts. The difference is not the number of models but who governs their use.