AI systems inventory
Which AI tools and automations exist in the company, who uses them and with what data. Including the ones built into other software.
Your company already uses artificial intelligence, whether anyone has approved it or not. We take the inventory, classify each use under the EU's AI regulation and put the rules in writing, with ISO 42001 if you want a management system.
Someone in sales pastes a contract into a free assistant and HR tries out a tool that screens CVs. Nobody decided it and nobody has the list.
The AI Act assigns obligations according to the risk of each use and to your role: whoever develops a system is responsible for some things and whoever uses it, for others. To know which apply to you, you need to know what AI you have.
This is the legal side: what AI is in your company, what risk each use carries and what applies to you given your role. The usage policy and team training are done once, under Safe AI and training, and here they form part of compliance.
Which AI tools and automations exist in the company, who uses them and with what data. Including the ones built into other software.
Each use measured against the regulation's categories: prohibited practices, high risk, transparency obligations and the rest. The decision is documented.
Approved tools, what data can be entered into each one and what is prohibited. Written for all staff to read.
The regulation asks organisations that use AI to take steps so their team learns to use it properly. We train your team on your tools and your rules.
If you have any, we prepare with you the human oversight, the logs and the documentation that apply to you given your role.
An AI management system with a policy, risks and controls, which can be certified. It shares its structure with ISO 27001.
A free one-hour session to see what AI you already use in the company and what worries you about it.
Scope, timeline and a fixed price, in writing. What is left out of the proposal is written down too.
Interviews by area and a review of tools. The result is a register of uses with their risk level.
A policy approved by management, a list of permitted tools and sessions with staff.
The inventory is updated every time a new tool comes in, and its risk classification with it.
Distribution, food, transport and the public sector. We handle their licence, implementation and version upgrades.
A Power BI dashboard that reads from the client's GRC platform: controls, evidence, risks and action plans by financial year.
One company's criminal compliance model, loaded into eramba: more than 70 risks and 100 controls, each with its owner and its review.
Data protection as a system that is kept up to date: processing, risks, processors and breaches.
Implementing your information security management system, from the risks through to the certification audit.
Your company's criminal compliance model, with its risks, controls and evidence on one platform.
Rules of use, approved tools and training so your team uses AI without exposing data.
Yes. The regulation also covers companies that use other people's AI systems. The minimum is knowing what is used, training whoever uses it and making sure no use is prohibited or high-risk.
The regulation is already in force and applies in phases. The prohibited practices and the duty to promote staff training already apply. High-risk obligations, such as those for recruitment, were postponed to December 2027. We go through the rest with you in the discovery session, based on the rules in force at the time.
The AI Act is a European law you must comply with. ISO 42001 is a voluntary standard that describes how to organise AI management within a company. Having it helps you comply with the law and prove it, but it doesn't replace it.
Both, for different things. The provider is responsible for the system meeting the technical requirements. The company using it is responsible for following its instructions, overseeing it and not using it for prohibited purposes.
Every day. SolveIT works with its own AI agents and has an internal policy setting out which tools are approved and what data can be given to each one.
If the answer is ‘more or less’, that's where we start. In an hour we'll see what you have and what the regulation requires of you.