Vol. 01 / 03
Memory.
Punched cards: data read by hand, one at a time.
Operational AI for critical environments
The answers are already in your data.
We know how to read them.
We build the systems that hold an organisation’s data together — plants, archives, business systems, sensors, documents — in a single operating model, on which AI agents execute decisions with an operator in command. For large enterprises, government and defence. They run on a dedicated cloud in Italy, or on self-contained machines installed at your own site: the data does not leave.
On-premise machines need no deep integration into your network: they stay self-contained, and integration goes only as far as you decide. We run them together with you — you do not need people in-house who already administer AI models.
7 days
to be effectively operational on your data
99.9 %
platform uptime
24/7
continuous monitoring, no exceptions
100 % EU
data stays in Europe. And stays yours.
Solutions
All solutionsFour capabilities, one system. No data leaves your perimeter.
The model that adapts to your examples stays yours. Paying by the token rents a capability that stays with whoever sells it. Why the value has to stay inside
The paradigm
AI-powered automation
for every decision.
Operational AI
How it worksFrom language to action.
Agents do not suggest: they execute. Every action runs against a written map of your machines, documents and processes, within the limits you approved. Every critical action goes through an operator.
data
ontology
AI agents
human operator
action
example · predictive maintenance
data
A press reports abnormal vibrations.
ontology
The system knows what it is: line 2, critical component, batch in progress.
AI agents
The agent proposes: slow the line, bring maintenance forward.
human operator
The operator reviews and approves.
action
The line slows down. The stoppage never happens.
The archive
Reading data is nothing new. Making it act is.
Vol. 02 / 03
Computation.
The first machines turn data into answers.
Vol. 03 / 03
Action.
Today agents read data and execute. The data stays yours.
Sectors
Where data becomes results.
Tap a sector to see an example.
Operational notes
All notes
Machinery with high-risk AI: the presumption of conformity is provisional
Since 27 July the Machinery Regulation’s presumption of conformity has covered high-risk AI too — but the AI Act standards it leans on are not in the Official Journal yet.
Medical devices: when the technical file and the paper leaflet actually expire
Regulation (EU) 2017/745 sets 2027 and 2028 for legacy devices; Regulation (EU) 2025/1234 extends electronic instructions for use.
Battery passport: what stays a trade secret from 18 February 2027
Regulation (EU) 2023/1542, Art. 77 and Annex XIII: the cell’s general chemistry is public, the detailed formulation is not. The line has to be drawn now.
From the group · SPLIDIA
- 9 September F and T are not EI: what a spec that only cites L.32 is missing ITU-T L.32 points to the American ASTM E814 test, with F and T values. Italy’s fire prevention code classifies penetration seals as E and EI, per EN 13501-2 and EN 1366-3/4. Two different yardsticks, no conversion table: what to put in the spec.
- 4 September Before the customer exists: what ITU-T Y.1564 actually tests ITU-T Y.1564 (02/2016) and Corrigendum 1 (06/2021) define an out-of-service test, run before delivery: which Tperiod was bought, why limits only bind up to the CIR, and the burst method left explicitly undefined.
- 4 September The board that outlives testing: the CCT in a fibre network contract Above the EU threshold, the technical advisory board must be set up within ten days of the start of works, not at the first dispute. And if a reservation is entered against testing, the board does not dissolve: it rules on that too.
The first step
Operational from week one.
A real use case, on your data, in production. Then it grows, week after week.
30 minutes video call free of charge
Start an operational trial It starts with a session with our engagement expert. Your data stays yours, always.