Artificial intelligence
We use AI in our products and in our work to save time and gain clarity, never to replace human judgement. And we do so under a safe-use policy built into our security management system, certified under the ENS MEDIUM category and ISO/IEC 27001.
Approach
Artificial intelligence is used in specific tasks where it adds value: drafting a clear report from the technician's notes, summarising statistics or spotting gaps in an inventory. In all of them, a person has the final say.
In our products
Real features, in use with our clients, that save work without taking away control.
The technician writes in their own words and the AI suggests a clear report for the client, which the technician accepts or discards; the original is always kept. It also suggests a fault diagnosis by make and model, reviews the quality of the fire protection inventory and comments on the monitoring indicators.
Find out more
Managers can request a summary of the job and safety form statistics, which appears on screen labelled as AI-generated analysis.
Find out more
Cameras with embedded computer vision count people entering and leaving at entrances, without turnstiles, to show the occupancy of each building and room in real time.
Find out more
We integrate generative AI into business processes through a layer that lets you switch provider: proposing objectives and indicators, suggesting the scope of a contract from the tender specifications while citing the text that supports it, or preparing surveys.
Find out morePrinciples
The AI proposes and a person reviews, accepts or discards. It is never the sole source of any critical decision.
In the application, AI-generated content is distinguished from what a person writes.
The AI receives the context the task requires, and never passwords, keys or other secrets.
We only use AI services that have been authorised after assessment within our security management system.
If your contract or tender specifications require it, AI features are not used in your service.
Any improper sending of information or manipulation of instructions is handled as a security incident.
In development
Our developers use authorised AI assistants to write, review and document code. What they propose goes through the same path as any other change.
On workstations, the AI assistants' configuration is enforced by the organisation and users cannot change it: for example, features that would send an entire conversation to the provider are disabled.
Safe-use policy
The use of AI is governed by a dedicated procedure within our information security management system, which applies to staff, collaborators and suppliers who use AI on behalf of ATISoluciones.
Frequently asked questions
No. We do not train models with client data, and the AI services we integrate are contracted on enterprise API terms, which do not allow the provider to use that data to train its models.
Yes. AI features are an aid, not a requirement: if your contract or tender specifications require it, they are not used in your service and the work is done without them.
External providers assessed and authorised within our information security management system. If your procurement requires it, we will tell you which provider is involved, what information it receives and with what safeguards.
Our AI features are supporting functions (drafting, summarising, analysing and counting) and do not make decisions about people. We apply the obligations on staff AI literacy and on transparency: generated content is labelled as such and is always reviewed by a person.
We use AI as a development assistant, but all code is reviewed by a developer before it is merged and goes through the same security scans as everything else, within our secure development process. Responsibility for the software lies with our team, not with the tool.
We help you integrate it into your applications with human review, assessed providers and the security of an ENS-certified management system.
