
AI automation · business processes
Process automation with AI:
starting from the work that repeats.
From the data. On 19 November 2026, among the websites of Italian accountancy firms I read, 22% has a client area for documents (22 of 100 verified). The sector’s data.
It starts from a step that is done by hand every day, measures what it costs, automates only that and looks at the number afterwards.
Real cases
- A real estate agency: an honest online valuationA valuation engine that starts from the official price data and says how wide the range is.
- A chat assistant for student residencesAnswers only from approved cards, faults as tickets, emergencies handed to a person first.
- AI per studi legali: il primo contatto (IT)An article in Italian: case intake, documents, calendar: the Radar data.
How I work: four steps.
- 01Diagnosis
I look at the processes and find where time is lost.
- 02Duel
I compare you with three competitors, point by point.
- 03Direction
I say where to start and what to leave for later.
- 04Demo
I deliver a working prototype to try.
Questions I am often asked
Which processes should be automated first?
The ones that repeat the same way and travel by email or spreadsheets: collecting documents, answering the same questions, quotes, reminders.
Does the management software need to change?
Usually not: the automation works alongside the tools already there and moves the data where it is needed.
How is the result measured?
With the same number before and after: time per case, requests left unanswered, errors. The Check 4D measures them at the start.
What about customer data?
It stays in the business’s systems; the automation sees only what it needs, with written rules.