Put artificial intelligence where it costs you the most.
In most companies AI came in without a process: people use it on their own, the results can't be checked and nobody is accountable for them. The engagement starts from how the company works today, finds the process that absorbs the most hours and puts it into production together with the people who will use it.
Featured in: SkyTG24, Rai, Il Sole 24 ORE, Forbes, Class CNBC, Corriere della Sera.
Four processes, up close.
What changes in a company when a process goes into production. The names and details are invented, the steps are not: they are the ones that keep coming up in the mappings.
The same quote, twice.
A returning client asks for an updated offer. On the left, the work as almost every company does it today. On the right, the same work with the system in production: the line items come from the history, and next to each one it says where it came from.
A question to the archive.
Contracts, email, CRM and meeting notes answer in plain English, with the source under every answer. Whoever asks doesn't need to know which system held the information.
A reply to a client, with the source beside it.
The draft cites the contract and the renewal behind every statement. Whoever reviews it checks the source in one click and sends it. No reply goes out on its own.
A meeting, with nobody taking notes.
The transcript becomes tasks with an owner and a deadline, tied to the production plan and to invoicing. The minutes exist even when nobody writes them.
Today: Open the ERP. Find the client's last quote. Copy the line items into a new file. Update the prices by hand. Rewrite the terms. Check nothing was left out. Send. After: Ask for a quote for that client. Send.
Company Brain, The 1596 Method: everything the company knows, in one place.
The name comes from 1596, the year the Italian Matteo Ricci wrote out the Western art of memory in Chinese for a governor whose sons were facing the imperial examinations.
The Company Brain is the company's single memory: contracts, quotes, email, ERP and meeting notes, indexed where they already live and searchable in plain English. It is the layer every process we put into production stands on.
Where the return shows up
Time
Quotes, proposals and replies start from the company's own history. Whoever writes them edits a draft instead of filling a blank page.
Cost
Every case handled takes fewer manual steps, so the unit cost drops without touching headcount. The math is done on monthly hours, where lost time almost never gets counted.
Errors
Answers draw on the real documents, not on what someone remembers of a contract signed two years ago. The source document stays next to the answer, so whoever signs can trace it back.
People
The people working the cases stop copying data from one system to another and spend that time on the hard ones. What stays manual is the non-standard work, which is exactly where a person is needed.
Why AI projects stall
Tools without a process
Everyone picked their own tool and uses it their own way, and nobody decided anything. The results can't be checked and nobody can say whether it's working.
Pilots that never reach production
Everyone likes the demo, then the project stalls. Almost always, nobody is there to integrate it into the systems the company already uses.
Scattered information
Quotes, contracts, support threads, meeting notes. The information exists, but it sits in ten different places and the AI can't see it.
Sector experience that speeds up the work.
Financial Services, Healthcare & Life Sciences, Manufacturing & Supply Chain, Professional Services, Retail & E-Commerce, Media & Entertainment.
Results in production.
Marilab (Diagnostics Group) (Healthcare)
Mid-size diagnostics group processing over 50,000 reports a month. Manual data entry and appointment scheduling were causing a 15% error rate and constant operational overload. -62% operating costs, +18% revenue. Analysis of clinical workflows and identification of three high-return use cases. Automation roadmap, vendor selection and implementation management: automated report processing, appointment scheduling and predictive inventory management.
Industrial manufacturer (NDA) (Manufacturing)
Manufacturer with €40M in revenue and a supply chain across four countries. Demand forecasting based on gut feeling. Inventory waste at 12% of cost of goods sold. -38% inventory waste, €1.2M saved per year. Mapping of the entire supply chain data flow. Demand forecasting identified as the highest-return use case. Six-week pilot, results validated against historical data, then the system extended to every plant.
Retail group (NDA) (Retail & e-commerce)
Chain with 25 stores and a growing e-commerce business. Customer service overloaded, average response time of 48 hours. Online conversion stuck at 1.8%. +34% conversion, -55% response time. AI-driven customer engagement system: first-level support assistant, personalized recommendations and predictive analysis of purchase behavior. Implemented in eight weeks with the development team.
How it works
Phase 1, About one week: Process mapping
We look at where the hours go: who does what, with which tools, how many times a day. The output is a list of the points where AI removes repetitive work, ranked by what each one is worth.
Phase 2, About two months: First project in production
We start from the top of the list. The tool gets chosen, connected to the ERP and the archives, and tested in the field by the people who will work with it. The old process is switched off once the new one handles the volume of a normal week.
Phase 3, Overlapping with phase two: Team autonomy
Training on the process that was built, documentation and an internal usage policy. The stated goal is that the company runs on its own when the engagement ends.
Who leads the project
Tommaso Maria Ricci, Italian entrepreneur, born in 1989. For more than fifteen years he has built brands and companies in Italy and abroad, always from inside the companies he founded.
Founder of Neverframe Inc., an AI video production studio based in Miami. Author of "Build Your Business from Scratch with AI", published on Amazon, and editor of the newsletter The Blind Spot Intel.
Interview in European Affairs Magazine, August 25, 2026.
How we work
One domain at a time
The first phase covers a single process, the one with the clearest return. The others follow once the first one runs and people use it without asking for help.
No vendor lock-in
The tool is chosen to fit the process. The choice starts from the systems the company already has, and if the right one is inside a subscription already paid for, that is the one we use.
An honest read on the return
If the mapping shows no return, the engagement stops at the mapping. The document with the points found and their priorities stays with the company.
Not legal advice
For regulatory obligations, from the EU AI Act to privacy rules, we work alongside the company's counsel. We bring the technical side and the documentation of how the system works. The legal opinion stays with counsel.
Frequently asked questions
How long does it take?
The mapping takes about one week. It needs two or three short sessions with the people who will use the system, without stopping the business. The first project reaches production in about two months, depending on how many systems need to be connected. The autonomy phase overlaps with the second one: training starts once the process is already running, so the team learns on something that works instead of on a slide.
How much does it cost?
The quote comes after the mapping. Before seeing how the company works, any figure would be made up. The scope of the first phase and its cost are agreed together, based on the points the mapping surfaces. A company that decides to stop there keeps the document with the points found and their priorities: it stays theirs, even if the engagement ends there.
Is my company too small?
The engagement makes sense when there is a process that absorbs many hours every week and repeats itself almost unchanged. Headcount matters little. A firm that handles cases in a continuous cycle has more room than a larger company working on jobs that are all different. The initial assessment makes clear right away which case applies.
Where does the data go?
Tool selection starts from the company's constraints, including those on data handling. The company sets the perimeter. When documents are sensitive we work with providers that process data in the region the company requires and do not train on customer data, or with models installed on the company's own infrastructure. The decision is written into the internal usage policy, together with what can and cannot be fed to the systems.
Do we need an IT department?
No, but someone in the company has to be responsible for system access. In companies without in-house IT, that is the outside provider who already looks after the ERP and email. The integration work is done by our team, and at the end we hand over documentation that an outside provider can read and maintain.
How is this different from a training course?
A course explains how to use a tool and leaves it to the company to apply it. Here the process is built and put into production before the training, and the training covers that process. A course is still part of the third phase, together with the documentation and the policy, because a system people don't know how to use gets abandoned within weeks and the work goes back to being done by hand.
Request an assessment.
The initial assessment identifies where the hours go in the company today and whether there is room for a first project.