Transformation through AI

Save

Several companies independently put similar numbers on the ratio: PAYBACK assumes 60 percent change and 40 percent technology, BSH 70 percent people and 30 percent technology.

A recurring diagnosis is that AI adoption is set up as an IT project and fails for exactly that reason, while the reported successes treat it from day one as a company-wide topic with HR and business units at the table.

Two companies report initiatives with high calculated benefits that nonetheless had to be put on hold because the organization and process maturity were not yet ready.

Status: confirmed, 5 new supporting episodes. The new episodes show the same finding. At Medtronic, Abby Rich observes that handing out AI tools brings neither a new way of working nor value. Mike Flynn of EY says that AI tools solve only part of the problem and that processes and organizational change are decisive. Moody's places its AI enablement function in the People team because Patrick Martin sees the adoption as a change initiative for people and not as a technology rollout. Christoph Knogler (KEBA) warns against delegating AI to IT, and Thomas Arnoldner sees the challenge at A1 not in the technology but in questions of culture and process. Dan Slagen reports that the difficulties at Zapier lay mostly on the human and organizational side. That the bottleneck shifts to actual use by people (Siobhan Savage, Dan Slagen) is already covered by the existing wording.

Supported by these conversations, 15 of them from the last three months

• Marketer Talk with Franziska KrippmarketerTALK · 22 Jul 2026 · 18 min · DE
  Spotify (listen with a subscription)Apple Podcasts (listen with a subscription)
• AI at Rossmann: Why Transformation Only Succeeds With PeopleKI Köpfe · 16 Jul 2026 · 34 min · DE
  Spotify (listen with a subscription)Apple Podcasts (listen with a subscription)
• Erste Bank AI Officer: If AI thinks for us, something went wrongAI Have a Dream · 02 Jun 2026 · 37 min · DE
  Spotify (listen with a subscription)Apple Podcasts (listen with a subscription)
• Tradition meets AI: how Bitburger is brewing the futureMarketing mit KI · 20 May 2026 · 49 min · DE
  Spotify (listen with a subscription)Apple Podcasts (listen with a subscription)
• Why AI Transformation Fails Without Culture: Genesys CIO and CPOTechnovation with Peter High · 13 Apr 2026 · 40 min · EN
  Spotify (listen with a subscription)Apple Podcasts (listen with a subscription)

Save

One corporation concludes after three years of using AI that data quality held it back, not the technology; industrial companies tie AI tools to cleaned-up master data and a unified ERP, and a skilled trades business calls digitalization the mandatory foundation.

Often the knowledge exists only in people's heads: NEUROLOGIQ has to make the maintenance technicians' knowledge of faults explicit, enovetic has so far failed on legacy systems whose business processes are in part known only to specialist users, and according to AI FIRST only the people themselves can write it down.

The dissenting view concerns timing, not substance: two organizations start without a finished data strategy and document as they go; according to the vendor meinGPT, companies that start right away also begin a cleanup project after a few months.

Status: confirmed, 4 new supporting episodes. Lennard Schmidt (Langdock) observes among customers that the value of AI depends on connected data, permission systems and documented processes. Christian Klein (SAP) attributes the 93 percent accuracy of a financial close assistant to around 100 finance systems, data silos and shortcomings in data quality, and Tiago Bianchi advises breaking down silos and checking the accuracy of the data beforehand. Dominic von Proeck (Leaders of AI) needed eight weeks rather than one hour of technical work for the standards of his LinkedIn agent, because they existed only in his head. The data silos (SAP, Bianchi) are one facet of the existing point on a unified ERP and data quality, not a new core.

Supported by these conversations, 15 of them from the last three months

• #30 BRACK.ALLTRON CIO Tobias Quelle: How strong AI use cases turn into a company-wide systemHUBER Business AI Podcast · 09 Aug 2026 · 42 min · DE
  Spotify (listen with a subscription)Apple Podcasts (listen with a subscription)
• Head of AI: When Does Waiting on AI Become a Mistake?AI & Why - Was verändert sich wirklich? · 29 Jul 2026 · 54 min · DE
  Spotify (listen with a subscription)Apple Podcasts (listen with a subscription)
• AI in the Trades: How a Well-Drilling Company Works with AI (Interview with Sebastian Ehlen)KI im Unternehmen · 11 Jul 2026 · 33 min · DE
  Spotify (listen with a subscription)Apple Podcasts (listen with a subscription)
• Tesa CFO Jörg Diesfeld: “We are all still underestimating AI”FINANCE Podcast · 02 Jul 2026 · 44 min · DE
  Spotify (listen with a subscription)Apple Podcasts (listen with a subscription)
• Newell Brands CEO Chris Peterson: Organization Transformation Through AIThe CPG Guys · 13 Jun 2026 · 46 min · EN
  Spotify (listen with a subscription)Apple Podcasts (listen with a subscription)

Save

The work step is rarely the bottleneck: for engineering change orders in aircraft manufacturing, Heusingfeld cites 8.5 percent value-adding lead time, and AI FIRST warns that quickly created content then waits two weeks for approval from a regional office.

Redesign comes before automation: Medtronic and Cisco examine which tasks can be dropped, EY designs processes under “Design for Zero” first without people, and Leaders of AI remodels processes instead of bolting AI on.

AI amplifies what the process already is: Max Kanat-Alexander sees every existing problem greatly magnified by AI and points to CI pipelines that do not work, and Stefan Atanassov warns that automation makes a bad process worse.

Status: sharpened. Two independent new episodes show where the gain seeps away: for engineering change orders in aircraft manufacturing, Alexander Heusingfeld cites 8.5 percent value-adding lead time and says AI alone only shortens the touch time. Max Kanat-Alexander says coding agents do not help when CI pipelines do not work or an elaborate human approval process stands in front of the work. Two independent episodes likewise show that AI magnifies the weaknesses of a process (Kanat-Alexander: AI as an amplifier; Stefan Atanassov: a bad process gets worse through automation). Medtronic, EY (“Design for Zero”), Leaders of AI and Magnus Baringer (working out processes with the practice team before digitizing) confirm that redesigning the process comes before automating it.

Supported by these conversations, 21 of them from the last three months

• If I Were Starting Over With AI TodayAI FIRST · 04 Sept 2026 · 36 min · DE
  Spotify (listen with a subscription)Apple Podcasts (listen with a subscription)
• What Happens When AI Adoption Actually Works? - with Eric Porres, Chief AI Officer of LogitechBeyond the Prompt - How to use AI in your company · 19 Aug 2026 · 65 min · EN
  Spotify (listen with a subscription)Apple Podcasts (listen with a subscription)
• EGYM CEO: "If you think AI will not change your job, you are in the wrong place here." (with Philipp Roesch-Schlanderer)AI FIRST · 07 Aug 2026 · 47 min · DE
  Spotify (listen with a subscription)Apple Podcasts (listen with a subscription)
• How Grant Thornton Reimagines Core Processes in the AI EraThe Data & AI Chief · 05 Aug 2026 · 43 min · EN
  Spotify (listen with a subscription)Apple Podcasts (listen with a subscription)
• How a Doctor Is Building the AI Operating System for 8,000+ Practices and Clinics with an AI-First TeamAI FIRST · 17 Jul 2026 · 57 min · DE
  Spotify (listen with a subscription)Apple Podcasts (listen with a subscription)