Service
AI consulting for companies: adoption strategy
Artificial intelligence consulting for companies and SMEs: where to use it, in what order and under what rules. Process map, priorities, step-by-step plan.
In brief
FT Support's AI consulting helps the people who run the company decide where artificial intelligence should deliver a result, in what order to tackle the processes, under what rules and with what budget. It covers the first steps of our method: a first conversation, a day on the process map and priorities, and a written plan in which every step has a cost and an expected result. The plan is the company's, to carry out with us or with others.
- The direction is set from the top; the use cases are gathered from the people who do the work.
- A process map with benefits, difficulties and available data.
- A step-by-step plan: each step with its cost, timing and expected result.
- Clear usage rules on tools, data and responsibilities.
- The decisions that fall to senior management, put in writing: objectives, priorities, responsibilities, budget.
AI consulting for a company: what it is
Artificial intelligence (AI) consulting is the work done with the people who run the company to decide where AI should deliver a result, in what order to tackle the processes, under what rules and with what budget, and to turn those decisions into a step-by-step plan that the company can carry out. An AI consultant does not sell a product: they look at the company’s processes and data, say where AI pays off and where it does not, and write a plan in which every step has a cost and an expected result.
It is AI consulting for SMEs and manufacturing companies first and foremost: companies where sales, purchasing and administration (accounts) spend hours searching for, copying and checking information, and where nobody can devote a whole department to artificial intelligence.
Why you need a direction, even in a small company
Artificial intelligence finds its way into a company anyway. Someone already uses an assistant to write emails, a supplier adds it to the ERP (the company’s management software), a department tries out a tool on its own. Without a direction the result is always the same: many scattered trials, confidential data copied into personal tools and no number to show whether it was any use.
An adoption strategy is a short set of decisions taken by the people who run the company: where AI should deliver a result, in what order, under what rules and with what budget.
The decisions that fall to the people who run the company
- The objectives. Two or three results that really matter: response times to customers, errors, margin, continuity of know-how when staff change.
- The priorities. Which process to start with and what can wait.
- The usage rules. Which tools are permitted, which data must not go into them, how results are checked, who the internal point of contact is.
- The responsibilities. Who looks after adoption and with what mandate.
- The budget and the criterion for continuing. How much to invest in the first step and which number decides whether to take the second.
- The people. Who will use the tools, under what rules, and how the company responds to what the AI Act requires on staff AI literacy.
These are choices that no department can make alone, and that no supplier can make in the company’s place. We bring the method, the examples and the knowledge of what the technology can really do today.
What we do, in three steps
- First conversation. About 45 minutes, no obligation. You tell us how you work and where time is lost; we tell you frankly where AI can bring a measurable advantage and where, today, it is not worth it.
- Process map and priorities. A day’s work with senior management and the heads of department, by video call, in English or Italian. We look closely at processes and data, collect the proposals of the people who do the work every day and put the processes in order, most worthwhile first.
- The step-by-step plan. A short document, the one many call a roadmap: the processes in order, most worthwhile first, and for each one what is done, how much it costs, how long it takes and what result is expected.
We ask senior management and the heads of department for little of their time: the first conversation, that one day and a short meeting to present the plan. The full path, through to the pilot project and production, is described on the Method page.
What you receive
- The process map, most worthwhile first, with the expected benefits and difficulties of each process.
- The data inventory: what there is, where it is, how far it can be trusted.
- The immediate improvements, achievable with ready-made tools.
- A proposal for usage rules, the so-called AI policy: permitted tools, data that must not go into them, checking of results, internal point of contact.
- The step-by-step plan, each step with its cost and expected result.
The plan is yours. You can carry it out with us, with others or in-house.
When we advise against starting
Saying so beforehand costs less than finding out later. We advise against a project when:
- the task comes up rarely, and the time it would save does not repay the work;
- the data needed does not exist, is not accessible or nobody keeps it up to date;
- nobody in the company can look after the project, not even for a few hours a week;
- the objective is vague, along the lines of “let’s use AI for sales”;
- the idea is to entrust AI with a decision that affects people, such as selecting or assessing workers, without the safeguards the law requires.
In many of these cases there is useful work to do first, usually on the data or on the description of the process.
Top-down or bottom-up?
There are two ways of bringing AI into a company. Top-down: senior management sets objectives, rules and priorities, then the projects start. Bottom-up: you start with individual tasks entrusted to AI assistants and agents, and what works is extended.
In our experience neither is enough on its own. What works is the combination: the direction and the rules come from the top, the use cases come from the people who do the work, and you proceed one process at a time. We cover this in the guide AI adoption in companies: top-down or bottom-up.
How to choose an AI consultant
The questions to ask anyone who offers artificial intelligence consulting, ourselves included.
- Which process would you start with, and why? Anyone who answers “all of them” has not looked at the company.
- What is the company left with at the end? A map, a data inventory, a written plan, or just a presentation.
- How is the result measured? A number before and a number after, or nothing.
- Who decides what AI can do on its own? The right answer is: the company, with written rules.
- Are you tied to a provider? If the answer is a product, the consulting is a sale.
- What happens if the pilot does not work? You must be able to stop having spent little and learned something.
- Have you put AI into production in your own company? Anyone who has not done so does not know the problems of a Monday morning.
Our answers are on this page, in the method, in the case studies and in the frequently asked questions.
Which companies it is for
For any company where skilled people spend hours searching for, copying, checking and re-entering information. We work with companies in Italy and across Europe.
With manufacturing companies and with those that buy, process or sell raw materials and metals we have an extra advantage: we know the sector, because since 2010 we have developed and run FT Mercati, and we know how a price, a price list or an indexation clause travels through purchasing, sales and administration.
After the plan
The first step of the plan is almost always a pilot project: a single process, real data, real users, a few weeks, one number to improve. Depending on the process chosen, it takes the form of an AI agent inside the ERP, an assistant on company know-how or custom software.
The plan is not written once and for all: after each step you look at the result and decide on the next, not least because the technology changes in the meantime. For companies in Italy, part of the investment that follows may qualify for digitalisation incentives, which we have collected in the guide Incentives 2026 for software and AI in companies (in Italian).
It is the path we have followed ourselves: the CRM with AI agents we use every day grew one process at a time. The other four services, that is, what gets built after the plan, are described on the services page.
Frequently asked questions
AI strategy and consulting: the most common questions
What does an artificial intelligence consultant do?
An AI consultant helps the company understand where AI brings a measurable advantage and where it does not, choose which process to start with and set its own usage rules. In our case the work includes the analysis of the processes with senior management and the heads of department, the priorities, and a written plan in which every step has its content, cost and expected result.
How does AI consulting with FT Support start?
With a first conversation of about 45 minutes, no obligation, in which you tell us how you work and where time is lost. If there is a process worth working on, we move on to a day with senior management for the process map and priorities, by video call, in English or Italian. If we do not find one, we say so.
How long does the work take, and what is the company left with?
The first conversation takes about 45 minutes, the process map with the priorities takes a day of work together, and the written plan arrives in the following days. The company is left with the process map, the data inventory, the immediate improvements, a proposal for usage rules and the step-by-step plan. The plan belongs to the company, which can carry it out with us, with others or in-house.
How is the price of AI consulting set?
By steps, not by the hour. Every step has its content, price and expected result written in a proposal before we start, and after each step the company decides whether to take the next. What the cost of an AI project for an SME depends on is explained in our guide on costs.
What does the AI Act require of a company that adopts AI?
Today, four things: not using prohibited AI practices, taking measures so that the people who use the systems know their limits and risks (the AI literacy of Article 4), complying with the transparency obligations and knowing whether any of the systems is a high-risk AI system. In the adoption plan these choices sit in the usage rules and in the responsibilities: who uses what, with which data, who is accountable. The full picture is in our guide on the AI Act for manufacturing companies.
Are company rules on the use of AI, the so-called AI policy, mandatory?
In general, there is no legal obligation to have a document with this name. It is, however, the simplest way to meet obligations that do exist: the AI literacy measures required by the AI Act, the protection of personal data and, in Italy, the information that national law requires employers to give workers in certain cases. For a company of fifty people one page is enough: permitted tools, data that must never be entered, checking of results, an internal point of contact.
Many people in the company already use ChatGPT on their own: where do we start?
With a no-blame survey: who uses what, to do what and with which data. It is the most useful picture you can have, because it shows where AI is really needed. Straight afterwards you need a few usage rules and company tools in place of personal accounts.
Do we need an IT department or an in-house expert?
No. You need a point of contact who knows the processes and can give a few hours a week. The technical side, that is data connections, security and development, can be outsourced. Decisions on objectives, rules and budget stay with the people who run the company.
Are you tied to a technology provider?
No. We mainly use Anthropic's Claude models, because we find them reliable in reasoning and in the use of tools, but we design our systems so that the model can be replaced. When a ready-made tool is enough for a task we say so, and that is often the case with the first improvements.
Besides the plan, do you also develop software and AI agents?
Yes. We build AI agents inside ERP and CRM systems, assistants on company know-how and custom software. So whoever writes the plan knows what can be built and at what cost. The plan can still be used with any supplier.
Do you have a process you would like to see working better?
Tell us about it in a first conversation, with no obligation: together we work out whether and where AI brings a measurable benefit, and which step to start with.