Insurance & asset management
AI pilots that reach the daily work, inside the AI Act and DORA
Your pilot works in the demo. It has never touched real work outside one team. The AI Scan finds out why, and which pilot is worth taking live.
The clock: Time until a pilot is in production, or stopped on purpose
Playbook: fictional demo
Four reasons a pilot stalls. The Scan checks each one.
Nobody owns the result.
The pilot belongs to a project team, not to the person who runs the process.
Security and legal see it late.
Their first look comes when the pilot is finished, and the answer is no.
Nobody signed what it is worth.
Finance asks what it saves, and there is no agreed figure.
AI is already in use without approval.
A note-taker or a chatbot, with no owner and no oversight.
Insurance · fictional
Muster, taken through the Scan
Muster Versicherungsgruppe is invented. The steps are the ones we would take with you.
Read the case- Discover Five processes, five AI systems. Two without stated oversight, one with no owner at all.
- Diagnose Ten dimensions with reasons: 46, value the weakest, five gates failing, each gap with a closure test.
- Prioritize Six use cases ranked by rule. Claims document intake is the only pilot. Two life and health cases are high-risk and wait. Fraud scoring waits for counsel.
- The readout A business case with a value floor and a named owner. The CFO and the COO sign go.
What the Scan settles
From the first call to Assurance: the register, the AI Act class, one pilot with a value floor, and who signs each step.
Insurance · fictional
- An AI register, including the tools nobody approved
- A preliminary AI Act class per use case; counsel decides
- One pilot chosen by rule, with a value floor the CFO signs
- Every condition from the CISO and legal turned into an owned control
What it does not do
- It is not legal advice. The AI Act class is preliminary; your counsel decides.
- It does not certify you against the AI Act or DORA.
- It does not build the pilot. That is the Last Mile, and only if the Scan says go.
- No insurer has run it yet. Muster is invented, and the badge on this page changes only with a paid Scan.
What changes in every industry
Know what you protect
Every Scan starts with the thing that must not stop: a line, a delivery date, a regulated service. Then it counts the days you have.
Owners, not dashboards
Every warning gets a named owner and a deadline. A tool that sees a risk nobody is bound to act on is a finding, not a control.
Proof the CFO signs
A value floor in your own numbers, signed before anything is built. Then scale or stop on what was measured.
Questions
Do you need access to our models or our data?
No. The Scan works from your register: use cases, systems and owners. If a pilot goes ahead, it runs on your own endpoint and key.
What if our pilot should stop?
Then the Scan says so, with the reason. A pilot stopped on purpose is a result, not a failure.
Is this an AI product?
It is a method with an engine. AI can shorten the days it takes to see a warning; a person always decides.
What data do you need?
Your own register: critical suppliers or AI use cases, the tools that watch them, and who owns them. A spreadsheet is enough to start.
What does it cost?
The Scan has a fixed price between €15,000 and €25,000, set by ten published drivers. You see the rule before you sign.
Has this been used with clients?
Every case on this page is fictional and labelled so. Each industry shows how far it is proven, and that only changes with a paid Scan.
Other industries
Automotive suppliers
The running line
Open the industry pageMachinery
The delivery date you promised
Open the industry pageElectronics
Your allocation
Next: not built yetStart with what you protect.
A short self-check first. Then, if it fits, a fixed-price Scan with the people who can say no in the room.