About Goose Group
Prepared for Aeaitalia.
Goose Group is a small, senior team that turns AI ambition into real operating capability. We work directly with leaders and operators to find where AI matters, build the first useful tools, and leave behind a way for the team to keep going on its own.
What we do
We work between three groups that often do not connect: the executives who set direction, the people who run the day-to-day operations, and the technical teams who build. Most companies already know that AI matters, and their teams can see where the work could be better. What is usually missing is not another software platform. It is senior product judgment, real fluency with AI, and the ability to build something that works and prove it with evidence.
Why this is not just advice
For AEA Italia, the relevant model is CoE + implementation. Goose Group brings the mechanism for finding and governing the opportunity, and the implementation capability to turn the strongest evidence into a working capability.
The first phase should still be controlled. The immediate goal is not to replace the current triage system. It is to measure how well current triage performs, identify where decisions are not optimal, and define the scorecard and test harness that would justify a larger program.
We have extensive experience in insurance
Alex and Mike worked together on a multi-year transformation at First American, a Fortune 500 US title insurance company. Over roughly five years, the programme rebuilt the software and processes behind the full insurance experience, including claims, underwriting, and customer service. The work was delivered through TheoremOne.
Alex Finnemore, Berlin
Alex has spent a decade leading AI and software transformation programmes for large, complex enterprises, often in regulated industries. At First American he helped rebuild core insurance workflows, including underwriting automation and customer service. He has led similar multi-year programmes for enterprises including American Express and Kia, working directly with executives to turn business-critical problems into systems that hold up in production.
He went on to senior leadership at Monks, a 7,500-person global technology firm, where he shaped its AI strategy and its partnerships with NVIDIA and Palantir. He holds a PhD in Physics from Cambridge.
Mike Joyce, Amsterdam
Mike spent five years at AWS, where he ran the innovation incubator for its largest global accounts, including Amazon, Meta, Netflix, and Salesforce. He sat with each business to find the highest-value opportunities to apply technology, tested them with data-driven experiments, and built the ones that proved out.
Earlier, he built the machine-learning strategy for Nike's data organisation and large operational platforms at AT&T. He is at his best solving ambiguous problems where the systems do not yet line up.
Felix Roux
Felix was the founder and CEO of Wonder, a video-conferencing tool used by more than four million people during the pandemic, which raised over EUR 12 million in funding. He began his career as a consultant at OC&C in London and McKinsey in Berlin, working on data analytics and financial modelling in healthcare. He holds degrees in Philosophy and Sociology from Oxford, LSE, and Harvard.
On this work, he provides the analytic framing that connects the claims diagnostic to a clear business case.
How we work
Start from the work. We sit with the people doing it and map how claims actually move: handoffs, systems, rules, judgment calls, and the places where context disappears.
Build the smallest useful thing first, and measure it against real data before anyone trusts it in production.
Make it governable. Ownership, logs, review, and clear decision points, so the system stays transparent.
Build ourselves out. We leave behind working capability, source, and playbooks so your team owns what we build together.
First phase
The sensible first phase is a read-only historical claims diagnostic for electrical mass claims. It should replay historical claims, measure current triage performance, identify recurring decision errors, quantify the operational impact, and outline the business case for a larger capability program.
The expected outputs are a current-state decision map, data feasibility assessment, first scorecard, test harness design, experiment plan, and recommendation for the next phase.
If the evidence supports it
The next phase would be a focused opportunity incubator around claims decision capability: data access, scorecard, test harness, prototype, first live operating area, governance, and business case.
This is a fee-based capability program. It should not be framed as an upside, JV, or venture model unless the commercial structure changes later.
Next step
We are glad to come to Italy to meet the team in person and walk through the approach. In the meantime, Philippe has the strategy brief, and we are happy to answer any questions from colleagues directly.