Artificial intelligence for the people who run a company.

Chapter 2

Where AI pays off first

Start where the work is repetitive, high-volume and already digital. In a typical SME the first returns come from five areas:

• Customer service: answering recurring requests, triaging tickets and emails, preparing replies for an operator to approve.
• Sales and marketing: qualifying leads, following up, drafting proposals and content from the company’s own material.
• Administration and finance: reading invoices and orders, reconciliations, recurring reports.
• Operations: demand forecasting, inventory, predictive maintenance and quality control on production lines.
• Knowledge work: finding and summarising internal documents, procedures and past projects.

Whatever the area, measure before you start: hours spent, error rates, response times, backlog. A pilot without a baseline cannot prove a return.

Chapter 3

A strategy that fits an SME

An SME does not need an enterprise AI programme. It needs five decisions taken in the right order.

1. One pilot with a measurable return within six months. Pick a single process, define the KPI and the owner, and set the go/no-go criteria in advance.
2. A roadmap, not a wish list. Sequence the next projects from simple automation to forecasting and agents, each building on the data and skills of the previous one.
3. Data governance. Decide where data lives, who can access it and which data may leave the company. On-premise deployment keeps sensitive data inside.
4. A cross-functional team. Operations, IT, HR and the legal or compliance function all sit at the table; the business owns the outcome, not the vendor.
5. A named owner for compliance and ethics. Someone keeps the inventory of AI systems, the documentation and the human-oversight rules up to date.

Budget realistically. A first structured project with PugliAI starts at €15,000 (AI Accelerator, three months); autonomous agents from €490 a month; on-premise products from €25,000. Subscription tools cost a few tens of euros per user per month, but they do not integrate with your systems on their own.

Chapter 4

Leading the change

AI transformation is cultural before it is technical. The leaders who succeed treat AI as a way to augment people’s work, not to replace it, and they say so explicitly.

• Involve the people who do the work in designing the automation; they know where the exceptions are.
• Address fears about accuracy, security and jobs openly, with facts and with the rules you have set for human oversight.
• Invest in skills at every level: since 2 February 2025 the EU AI Act (Article 4) requires companies that use AI systems to ensure a sufficient level of AI literacy among their staff.
• Celebrate the first measurable wins and share what did not work; both build trust.
• Keep a person accountable for every automated decision that affects customers or employees.

Chapter 5

Regulation: the EU AI Act and Italian law

Regulation (EU) 2024/1689, the AI Act, applies in stages. Most SMEs are deployers of AI systems rather than providers, which keeps their obligations manageable: know which systems you use, tell people when they interact with AI, keep human oversight on decisions that matter, and comply with the high-risk rules only where a system falls into those categories.

In November 2025 the European Commission proposed a “Digital Omnibus” that would postpone parts of the high-risk regime, with backstop dates in late 2027 and 2028. At the time of writing it is a proposal, not law: plan on the dates above and check the current status before you commit budget to compliance work.

Italy’s Law 132/2025, in force since 10 October 2025, is the first national AI law in the EU. It designates AgID and ACN as national authorities, requires that workers be informed when AI is used in the workplace, sets rules for AI in healthcare and the professions, protects works with a genuine human creative contribution, introduces a criminal offence for the unlawful dissemination of AI-generated or altered content, and authorises an investment programme of up to €1 billion through CDP Venture Capital.

A compliance framework that fits on one page

1. Inventory and classify: list every AI system in use and assess its risk level.
2. Document: keep records of data sources, suppliers, purposes and decisions.
3. Human oversight: define who reviews and can override automated decisions.
4. Transparency: inform customers and employees when they interact with AI, and label AI-generated content where required.

GDPR continues to apply in full: lawful basis, data minimisation, impact assessments where needed, and processors under contract.

Chapter 6

Your 90-day action plan

Ninety days are enough to go from intention to a running pilot.

1. Assess your starting point (weeks 1–2): processes, data, skills and the tools already in use.
2. Pick the use case (week 3): clear return, short implementation, an owner who wants it.
3. Fix the data (weeks 3–6): where it lives, who can access it, what may leave the company.
4. Train the people involved (weeks 4–8): how the tool works, its limits, the oversight rules.
5. Set governance (week 6): the AI inventory, documentation and the person accountable.
6. Use the ecosystem (ongoing): consultants, universities, incentives such as Transizione 5.0 and regional grants.
7. Measure and iterate (from week 8): KPIs against the baseline, then scale or stop.

AI is a management decision before it is a technology decision. With a clear first project, honest measurement and a team that understands the rules, an SME can turn it into a lasting advantage. PugliAI is at your side on this path.