> ## Content Index
> Fetch the complete content index at: https://www.tippinpoint.ch/llms.txt
> Use this file to discover other available public pages before exploring further.

# In, On or Out of the AI Loop
- URL: https://www.tippinpoint.ch/in-on-or-out-of-the-ai-loop/
- Published: 2026-11-20T06:19:42.000Z
- Updated: 2026-11-20T06:19:42.000Z
- Description: There’s a long queue on Monday morning to enter the Kunsthaus in Zurich. The museum is hosting the Swiss AI Summit. Jeffrey Gibson’s imposing sculptures in the foyer, with their bright colours and gold, evoke change and transformation. An apt backdrop and welcome.
- Author: Vega Ibanez
- Tags: Tech, Swisscom, #autorenbild, #autorenbox, #pos4

The crowd is senior and institutional. Big banks, large corporates, government institutions and a few startups. My friend, who also attended last year, sums up the shift over coffee: “Last year was about why. This year is about how.”

Swisscom explains the how. Staff can develop their own AI use cases and AI agents. A slide sorts their current projects. About 20% assist: speeding up work. About 70% redesign and automate. Only about 10% create something that was not there before. 

![](https://storage.ghost.io/c/11/7a/117a3396-8af6-43a4-8dc5-f891a0cee13d/content/images/2026/11/unnamed--19-.jpg)

Only about 10% create something that was not there before. Foto: Vega IbanezV

It’s less clearly explained if one project in ten doing something new is fast or slow innovation. But it makes sense that the people closer to the process are allowed to improve and automate their more unthankful tasks, opening time for more valuable tasks.

An anecdote told at the coffee break tells the story differently. An unnamed company went looking for its biggest token spender, probably hoping to find a new efficiency application to roll out to everyone else. But the tokens were being used to run a Tinder agent. The innovation was an autonomous matchmaker that would swipe left or right according to its master’s rules of beauty. Not all employees will choose the right problem for the company paying the tokens. 

What is clear is the corporate push to continue build agents. But, once we have “agentized” everything we don’t like doing, when we have applied AI to all levels of corporate inefficiency, what happens to the staff?

Maybe we can become supervisors of our own creations. A subsequent panel explains how AI governance already requires humans in the loop, on the loop, or out of it: clicking to approve, watching and stepping in when something goes wrong, or reviewing afterwards what the agents did.

Rather than mid- or senior managers, instead of Vice President, Director or Executive Director, we will be specialists in In, On or Out. 

The Summit is closing with music and chatter in the grand foyer. I leave, and Rodin’s Thinker still sits at the top of the Gates of Hell looking down at the mayhem in bronze. I salute him silently and head home in the rain.