Recap: AI Solution Day 01, Customer Care
Eight real challenges on the table, and the three questions that came up at more than one of them.
On 24 March we ran the first AI Solution Day at the hub in Praterstraße 1. Five hours, from one in the afternoon until six, with startups, scaleups and corporate innovators in the same room.
The format was simple. Startups pitched, then everyone sat down together to work on eight real challenges brought in by corporates.
What did not happen was the usual thing, where a room disperses into business card exchange the moment the pitching ends. Instead: mixed groups, real problems, and a dynamic that turned into a group effort to understand what was on the table and how to solve it.
Some follow-up meetings were booked before the day was over.
What surfaced
Three questions came up across more than one table, which is usually the sign that something is a real problem rather than one company's problem.
Measuring the ROI of AI in customer service. Easy to demonstrate, hard to prove. Handling time drops, but so does first-contact resolution if the deflection is poor, and the two effects hide each other in the aggregate numbers. Several groups landed on the same conclusion: you have to instrument the escalation path before you deploy anything, or you will never know what the system did.
How human should a bot be. The room split on this, and the split was not between startups and corporates. One argument is that users prefer a clearly labelled machine and resent being made to guess. The other is that warmth measurably improves resolution rates for frustrated customers. Nobody resolved it, which is itself useful to know if you are about to make that decision for your own organisation.
Corporate email volume, without losing the plot. The most concrete of the three, and the one where startup solutions and corporate reality were furthest apart. Triage and drafting work well in demos. What kept coming up was accountability: who is responsible when an automated reply is wrong, and how you keep a human meaningfully in the loop without recreating the original workload.
Why the format works
The pitching is the short part of the day, and it is not the point. The long part is groups of people sitting at tables with a problem in front of them, where the corporate has to explain their constraint properly and the startup has to stop pitching and listen.
That is uncomfortable for about ten minutes and productive for the rest of the afternoon.
Thank you
To our co-hosts GlassDollar and AI Austria, and to everyone who brought both their energy and their real challenges into the room. The challenges are what made the day work. A roundtable with a hypothetical problem is a workshop. A roundtable with a real one is a meeting.
What came next
Edition 03 takes the same format to HR and people, on 20 October 2026 at the hub. Ten startups pitch, then corporates work through the hiring, development and retention challenges they have submitted in advance.
Programme and applications for the current edition at aisolutionday.at.
If you have a challenge worth putting on a table, get in touch.
