Large organisations are using AI, but few have redesigned their core operations around it, Accenture CTO Marc Warner told a Fortune panel. The gap, he and other executives argued, is less about adding another tool than rethinking how work gets done.
Accenture AI Watch analysis
What happened
Warner said that organisations “really, truly reinvented with AI at the absolute core” are “basically close to zero”. He described common uses such as marketing, meeting notes and transcription as peripheral, while saying coding is a more measurable exception.
The discussion also featured executives from AXA, Mastercard and Mistral. Mastercard’s Greg Ulrich said speeding up one step in a long process can simply create a backlog at the next; the harder task is remapping the workflow. AXA’s Matthieu Caillat described narrowing an initial pool of about 400 AI use cases. AXA’s stated target is €500 million to €700 million in annual recurring pre-tax value from AI by 2029.
Fortune says Accenture was a founding partner of the event and sponsored the session. The article reports that its journalist independently verified the factual claims.
Why it matters
The panel’s useful distinction is between adopting AI and changing a business around it. Faster drafting or meeting notes may save time, but they do not automatically change the economics of a whole operation. A faster step in an unchanged process can merely move the queue along.
That makes workflow design, management and staff involvement part of the AI story, not administrative tidying after the technical work. The executives’ examples are company accounts, not proof that every organisation should follow the same playbook; AXA’s financial target is an ambition, not a realised result.
Our read
“Put AI in the workflow” sounds tidy until someone asks which workflow, who has to change it and what happens to the other 79 steps. The panel makes a persuasive case that the hard part is often organisational rather than a missing model. For anyone weighing an AI project, start by mapping the work and its bottlenecks, not by counting pilots.
What to watch
- Whether companies publish measurable results from redesigned workflows, rather than adoption totals alone.
- Whether AXA reports progress against its 2029 value target.
- How organisations handle responsibility when AI systems fail, especially in security-critical incidents.
Discussion spark: Should companies prioritise redesigning whole workflows around AI, even when that means disrupting established roles and processes, or is gradual adoption the more responsible route?
Sources and evidence
not affiliated with or endorsed by Accenture AI