Application, Ramona Furter
Application, Ramona Furter
AI Solutions & Delivery Manager, Lausanne
The first reviewer is an AI I built at Swiss Post, where I lead AI-driven business models from opportunity sizing to launch and own the build-vs-buy and cost calls that come with them.
Below: one incoming AI request at IMD, answered three ways, with the part most business cases leave out.
Your ad makes pruning and merging AI tools a core, ongoing accountability and puts cost stewardship next to it. Your centre already publishes a GenAI map that drops dormant tools on its own. Here is that same discipline applied inward: one request, answered three ways, judged on what your team still owns a year later.
Worked example, illustrative figures
"Proposals for custom programs take too long to turn around. Can AI help us draft them faster?"
Worth answering well: custom programs were 53% of operating revenue in your 2024 annual report, across 372 programs.
Three honest ways to answer it. Which one would you fund?
What it is
A proposal template and prompt library plus a reviewer checklist, built in Copilot Studio on the Microsoft 365 Copilot you have already given every member of faculty and staff.
Time to first use
Days. Your own Microsoft story describes shipping a dedicated Copilot to colleagues in Singapore in four hours, so the capability is already in the building.
What your team still owns at month 12
It answers the actual complaint, drafting time, on licences already rolled out across the school, and it adds nothing to the estate you are being asked to simplify. If it does not work, you know inside two weeks and you have spent two weeks.
Fund option 2 when the library is used on most custom-program proposals for two months running and reviewers still lose real time hunting for past material. I would write that number down before starting, not after, so the decision to spend more is already made when the evidence arrives.
At Swiss Post I lead AI-driven business models from opportunity sizing to a prioritised roadmap with KPIs, and I build the working prototypes myself, which is why the Swiss Post numbers above are mine to defend. Outside work I run a live product with real users on serverless cloud and public APIs, including a published iOS app, so releases, monitoring, support and the monthly bill are not abstractions to me.
The AI reviewer at Swiss Post got used because it sat inside the work people already did and left the sign-off with them. Before that I led a cross-functional team and external agencies at Ifolor on a CHF 100M+ business, and the marketing and sales team at WePractice after its Series B. In a school where faculty and business units each have their own way of working, that last mile is the whole job.
Every option in the case above carries what it costs to keep, because that is the number that decides whether a pilot survives its second year. At Swiss Post the build-vs-buy and cost-vs-benefit calls are mine, and on my own product I pay the monthly bill myself, which is a fast teacher about consumption pricing.
Before proposing anything new, I would find out what is already running and what it costs, because that is the half of this job the ad names twice.
Every AI tool, pilot and licence in use across the business units, with its owner, its cost and one honest line on what it changed. One intake route in, in line with the PMO practices your ad names.
Three or four decisions taken and communicated across the business units, each with the measure that would change my mind. Where two units have built the same thing twice, one of them wins and I go and explain why in person.
The highest-value use case shipped with human approval points and a named owner, running cost tracked from day one, and the AI Advanced team working the same way across Lausanne and the Cape Town hub rather than one time zone at a time.
AI business models from opportunity to launch; build-vs-buy and cost calls; prototypes for C-level.
IT, software engineering and AI services company.
Live platform with real users, built and operated end to end, plus a published iOS app.
CHF 100M+ business, cross-functional team and agencies, +9% conversion.
Partnerships with UBS and Baloise; bridge between client stakeholders and the product team.
Two funding rounds, 10 locations, 23 people; led the team after the Series B.
Go-to-market for internal startups, validation to scale-up.
Market pilots from MVP to launch: Smide, XpertCheck, Lizzy.
Promena, Cruspi, Domaco, Kuoni and AMAG; commercial apprenticeship at Bridgestone Switzerland.
Your centre ranks the world's 300 largest companies on AI maturity, and your faculty teach executives how to run exactly this transition. That makes the internal work in this job unusually consequential, because it is what keeps the teaching honest. Most places I could do this work would treat the AI estate as an IT line item. Here it is part of the argument the institution makes for itself, and that is worth moving for.
I designed and built this page with Claude Code, the same toolchain I use at work. The research, the assumptions and the recommendation are mine.