For the business
The few large calls a year, where one mistake is expensive.
- Capital investments
- Market and site expansion
- Pricing changes
- Partner and supplier selection
NewBusiness Decisions
Decisions grounded in your legacy, with the odds stated.
Calibrated probabilities
Each probability is checked against what actually happened to your past decisions, so 0.7 means seven in ten, on your record.
Cited precedents
Every figure names the past decisions it rests on and how many there were. Nothing is asserted without its sources.
Abstains when history is thin
Where too few precedents exist to state odds honestly, it says so and routes the decision to a person instead of guessing.
For your business, and inside it.
Inside the business
The many routine calls a day, where consistency is the gain.
- Approvals and exceptions
- Credit limits
- Inventory and reorder levels
- Staffing and shift planning
How it works
From your record to a decision you can defend.
- 01
Read the legacy
Past decisions and what came of them, drawn from your ERP, CRM, finance and the documents around them. Cleaned, linked, and kept inside your perimeter.
- 02
Weigh the precedents
For a new question, the situations most like it are found in your history, and their outcomes become the evidence, weighted by how alike they are.
- 03
Return a structured decision
The options, each with a probability of return, the range around it, and the precedents behind it. Or, where the evidence is thin, a handover to a person.
What comes back
One brief per question. Every option carries a probability, the range around it, and the number of past decisions behind it. Where there are too few, it says so and hands over.
Decision brief
Illustrative · not client data| Option | Probability | Precedents | Verdict |
|---|---|---|---|
| AOpen in Q1n = 23 | 0.680.61–0.74 | 23 | Recommend |
| BOpen in Q3n = 9 | 0.410.30–0.52 | 9 | Consider |
| CLease partner capacityn = 2 | Insufficient precedent | 2 | To a person |
Guardrails
Odds, never certainty.
- Probabilities calibrated on your own outcomes
- Every figure cites its precedents
- Abstains and hands over when history is thin
- Zero data retention by the model provider
- EU-hosted in Turin, GDPR compliant
- A person signs off every decision
Questions
How much history does it need?
Enough comparable decisions with known outcomes. Where a type of decision has too few, the brief says so and abstains rather than inventing odds. The pilot starts by measuring exactly this for your data.
Which systems does it read from?
Usually your ERP, CRM and finance systems, plus the documents where decisions were argued and approved. Ingestion and cleaning are part of the setup, as with the Knowledge Base.
Does it make the decision?
No. It returns options, odds and evidence. A person decides, and the record of who decided what becomes part of the history the next brief learns from.
Where does our data go?
Nowhere it is kept. Models run under enterprise zero-retention terms or in isolation, hosted in the EU, and your history stays in your own infrastructure.
How long is a pilot?
A pilot covers one decision type end to end: ingesting its history, calibrating against past outcomes, and running live briefs alongside your team.
Bring one decision. See its odds.
A pilot takes one type of decision you make again and again, reads its history, and runs live briefs alongside your team.
Decision support, not financial advice. Probabilities are estimates from your own historical outcomes; the decision stays with your people.