Evolution Exchange Nordics Podcast

The episode has scattered genuine moments - the Excel-sheet heuristic as a readiness signal, the 'sprinkling AI seasoning' sub-optimisation framing, and trust as a design problem rather than model-performance problem - but these are surrounded by large stretches of filler, acknowledged clichés ('low hanging fruit, I know that's been said a thousand times'), and generic change-management platitudes about mindset and communication.

“Look for Excel sheets, and in particular look for really complicated Excel sheets...It's practically an application in and of itself. That's usually a signal that, that is ready to be turned into a process”

“the pattern I'm seeing, uh, in general across big companies is we're sprinkling some AI seasoning on all the existing stuff, trying to optimize those specific tasks. But then we're sub optimizing whole”

A couple of non-obvious claims surface - trust as a design problem, the 10x productivity skepticism anchored to where project time actually goes, and the design-vs-execution breadth/depth split - but the bulk of the conversation rehashes standard change-management and AI-adoption advice found in any enterprise AI playbook.

“For me, trust, it's not a um, model performance question, uh, at least in my experience it's a design problem”

“the 10x that the big tech companies promise us, 10x or 10 times improvement of the amount of work that you can do is simply not true. Because I think that AI definitely speeds up a lot of processes...but I think that the entire project lifespan has not drastically decreased”

All four guests are genuine practitioners at recognisable organisations (Volvo Group, Ericsson, Vattenfall, and a 45-person AI consultancy) who speak from hands-on implementation experience, which is worth credit; however none are at C-suite or VP level and none have clearly led AI transformations at exceptional scale, limiting how much depth they can authentically offer.

“I'm Alan Smith, I work in design and product leadership at Volvo Group and I'm focused on complex internal enterprise products and how we integrate AI into new human agent interactions”

“on one hand I work for the innovation department and on the other hand I lead a team of data scientists trying to implement AI solutions”

The episode relies heavily on abstraction and hypotheticals; the few concrete anchors that appear - Vattenfall's ~20,000 employees, the fax-machine-used-three-times-a-year illustration, a vaguely cited Amazon deployment incident - are either trivial or lack verifiable detail, and there are no named AI projects with outcomes, timelines, or financial metrics.

“if you're going to do an AI project for swapping out a fax machine, probably the fax machine is still there because this workflow gets used three times a year”

“they put in a process where all check uh ins have to be human reviewed now by, by a senior engineer”

The host is a recruiter who functions almost exclusively as a turn-taking moderator, asking no substantive follow-ups and concluding segments with lines like 'has anybody else got anything else they'd like to add'; the more interesting exchanges are guest-initiated and even those rarely push back hard on unsupported claims, leaving speculation about layoffs and 10x productivity unchallenged with data.

“And has anybody else got anything else they'd like to add on to that?”

“Is there any topics anybody, or question, anything that anybody would like to revisit?”