AI usage policy
A short document saying what can be done with AI, what can't and who authorises a new tool. Management approves it.
Which tools can be used, what data they can be given and who is responsible for the result. We put it in writing, set it up and train your team.
With most assistants, free and personal accounts can use what people type into them to train their models. When someone pastes a contract or the client list in there, that information has left the company.
Banning AI doesn't fix it: people keep using it, just on the quiet. It works better to say clearly which tools are fine and what they can be given.
It's what we have written down and apply at SolveIT, adapted to your company. The legal side of the regulation (the AI inventory, risk classification and ISO 42001) is covered under Responsible AI and the AI Act.
A short document saying what can be done with AI, what can't and who authorises a new tool. Management approves it.
Which assistants can be used with company data, always with a corporate account. Anything not on the list isn't used for work.
A simple table, from public information to confidential. For each level, whether it can be used and under what conditions. Passwords and keys, never.
AI helps and the person signs off. Anything that goes to a client under the company's name is reviewed first by whoever is responsible for it.
Hands-on sessions with the approved tools and with cases taken from each team's work, because not everyone uses AI for the same thing.
Sign-in with company credentials and, if needed, blocking of unapproved assistants and of file uploads from company devices.
A free one-hour session. We look at which tools your team already uses and with what kind of information.
A draft of both, adapted to your company, which we review with you and management approves.
The chosen tools set up with company accounts, and sign-in with people's usual login.
Sessions by team with the approved tools, with exercises based on their real day-to-day work.
Tools change quickly. Every year we go over the list, the policy and what the team needs to learn.
One agent keeps the accounts up to date inside Odoo, another coordinates projects and another answers with what the company already knows.
A phone and WhatsApp assistant that answers questions, books appointments and hands over to a person. Under way, in phases.
An AI assistant that runs on a group-owned machine, with cloud models, skills for each company and a private area for each person.
Agents connected to your ERP, support desk and email that answer with your data and leave actions ready for you to review.
Your email, documents and ERP searchable in natural language, with answers that cite their source.
We connect your software with n8n and Make so data moves from one system to the next on its own, with alerts if anything fails.
An inventory of the AI your company uses, risk classification, rules of use and a management system under ISO 42001.
Those with a business tier and terms that rule out using your data to train models. For example, Claude, ChatGPT, Microsoft Copilot or Gemini on their business plans.
Passwords, keys and any other credentials. Nor non-public financial information or anything covered by a confidentiality agreement. Personal data only in tools with a data processing agreement and processing in the European Union.
Yes. It bans certain practices for any company that uses AI and asks it to take steps to train its staff. If you use it to make decisions about people, as in recruitment, the obligations are greater. We cover the detail from our GRC line.
Hands-on, with examples from each team's work. We adjust the format and length to each group.
Yes. At SolveIT we work with an AI usage policy approved by management, with its list of tools and its data classification. The one we propose to you starts from there.
Tell us how you work today. In an hour we'll see what rules you need.