
When two systems give two different answers, you start with the data
It happens that the ERP and the CRM show two different values for the same customer. We connect the sources and fix shared definitions, so the analyst, management control and an AI agent read the same number. The work the company has already done on data governance, such as definitions, owners and quality rules, is the foundation on which we prepare the data for artificial intelligence.
Customer
- Definition
- Shared by all functions
- Owner
- A named person
- Quality
- Measured
- Who can read it
- Authorised AI agents, by purpose
- Gestionale ✓
- CRM ✓
- Reclami ✓
“How many active customers per branch?”
- Source
- Approved indicator
What it contains
Data Foundation prepares the data on which everything else is built: from analysis with Data Intelligence to the AI agents working in processes.
01
Connectors and integration
They connect ERP systems, CRM, core banking and document archives, without replacing them and without migrating data
02
Quality and traceability
They measure completeness and freshness and, through data lineage, trace where each piece of data comes from and how it was transformed, so those who use it can verify it
03
Shared definitions
A single glossary for all functions: what counts as an “active” customer, when “time to decision” starts running, when a case is “complete”
04
Data product
Data becomes a product, with an owner, a usage sheet and quality measured against a threshold, ready for those who have permission to use it
05
Governed access
Sets which data an AI agent reads and for what purpose, under the data protection rules that apply to the company
The questions we ask of data before giving it to a person or an AI agent
A recognised source
It is known where it comes from and who updates it
A shared definition
It means the same thing to every function
An owner
One person answers for it and decides how it changes
A quality rule
Completeness, accuracy and freshness are measured
A permission
It is settled who, person or agent, can read it and for what purpose
A metric that finance and the network calculate the same way
What happens if margin per branch comes out in two different versions, one from finance and one from the sales network? We establish a shared definition of the metric together with the finance function, which becomes its owner. From then on the management dashboard, the network and the AI agents read the same number.
- Margin per branch comes out in two versions
- We establish a shared definition with the finance function
- The management dashboard, the network and the AI agents read the same number
Illustrative example
Frequently asked questions
Do existing systems have to be replaced?
No, we work with what is there and connect what the process needs.
Do we have to put all our data in order before we start?
No, and it is rarely worth it. We start from the data the first use case needs and prepare it to be reusable, so the next project that uses the same data finds it ready and costs less. This is the logic we call Compounding Intelligence.
Do we need a data strategy first?
The strategy says where to go; the use cases are the steps that get you there. If the direction is already clear, you start from the first use case. If there are many unconnected initiatives, it is worth investing time in defining the strategy first, so that all the effort goes in the same direction.
How does it differ from Data Intelligence and the Company Brain?
Data Foundation prepares the data, Data Intelligence analyses it and derives forecasts, and the Company Brain keeps the rules and the decisions taken on those numbers.
Understand where your data stands today
In a first meeting we look together at the data behind a process you care about and tell you where it makes sense to start.