AI Advisory

When it is needed

For companies that feel they must “do something with AI” but do not know where to begin. I start from work processes, not from tools.

What I do in practice

• Analysis of how work is done today and of repetitive or low-value activities
• Identification of the cases where AI brings a concrete advantage and of those where it does not pay off
• Assessment of costs, benefits and difficulty for each case
• A list ordered by priority, with the first steps to take

When it is needed

For companies that already use AI tools, even without having officially chosen them, and want clear rules before they become a problem of data, liability or compliance.

What I do in practice

• Survey of the AI tools already in use in the company, including those adopted independently by employees
• Definition of simple rules of use: which data can be entered, in which tools, with which checks
• Assessment of the risks for personal data, confidentiality and reliability of the answers
• Check of alignment with European AI regulation and personal data protection (AI Act and GDPR), highlighting the points to keep under watch

When it is needed

For those who must choose among different AI solutions and want to avoid decisions driven by vendor marketing.

What I do in practice

• Definition of what the company really needs
• Comparison of tools and suppliers on features, costs and where the data is processed
• Attention to contract terms and to the risk of depending on a single supplier
• A reasoned recommendation, with the alternatives assessed

When it is needed

For companies that want to reduce manual work on repetitive activities such as requests, documents, purchasing and supplier management. I proceed in small steps, checking the results before extending.

What I do in practice

• Choice of a limited activity on which to run the trial
• Design of the flow: what AI does, what stays with people, where human control is needed
• Building a pilot and measuring the results (time saved, errors avoided)
• Guidance, based on the numbers, on whether to extend it, correct it or stop

When it is needed

For companies that want AI to be used well by those who work every day, not just by a few experts.

What I do in practice

• Training sessions on what AI can and cannot do, with examples taken from everyday work
• Practical guidance: how to phrase requests and how to check the answers
• Training tailored to the role (management, operational offices)
• Support during the first weeks of use, collecting doubts and problems

When it is needed

When a single assistant is no longer enough: work runs through several steps, tools and data sources, and several AI agents need to work together without anyone losing control of what they do.

What I do in practice

• Breaking the process down into tasks, and choosing which to entrust to an agent and which to leave to people
• Design of the architecture: agent roles, the tools and data each one can access, how work is handed over
• Definition of the human control points and of the traceability of every step
• Building a first working flow and measuring the results before extending

When it is needed

When knowledge is scattered across documents, emails, notes and people’s memory, and finding a piece of information or a past decision takes longer than it should.

What I do in practice

• Survey of the sources: what exists, where it lives, who keeps it up to date
• Design of the structure: how notes, documents and decisions are organised and linked
• Building the knowledge base and connecting it to an AI assistant that queries it and cites its sources
• Rules for updating and access, so that it stays reliable over time