AI-Wiki

Is Agentic AI upending the corporate ladder? EY's Global Consulting AI Leader shares what’s coming

13 min read

Diasio & Wood (WorkLab) — agentic AI and the corporate ladder

▶ Watch on YouTube · Microsoft · 29:55

EY’s Global Consulting AI Leader Dan Diasio joins host Molly Wood to discuss how agentic AI is shaping the future of work and how companies can reinvent themselves as AI-first Frontier Firms. Learn how AI is disrupting the consulting industry, why investing in upskilling your employees is now mandatory, and why the right combination of mindset, skill set, and tool set matters more than ever. Listen in to discover practical strategies for leaders to drive transformation and compete with AI-native disruptors.

TL;DR

A 30-minute WorkLab episode on the Microsoft channel, published 16 December 2025. Host Molly Wood interviews Dan Diasio, Global Consulting AI Leader at EY. Wood introduces EY as a member of the inaugural class of the Frontier Firm AI Initiative, which Harvard runs with Microsoft (see HBS AI Institute). Five claims:

  1. AI raises the floor, not the ceiling. EY has run hands-on sessions with management teams “hundreds of times”, in which each team uses AI to invent a snack product. The results converge: “it’s all the same looking stuff… we would have a global shortage of matcha and monk fruit.” What raises the ceiling is “AI and something”: creativity, critical thinking, systems thinking and deep domain expertise.
  2. Productivity has a limit; growth does not. Success “takes the mindset, the skill set, and the tool set,” and “most of the investments have gone towards the tool set.” He sees a shift “away from use cases towards more of a reinvention agenda.”
  3. Write the strategy down, then build to a blueprint. First, a two-to-three-page point of view “you would be comfortable taking to your investors.” Then each function is rebuilt in layers. The top one is a control layer that decides how much autonomy agents get and where they are registered.
  4. The corporate ladder is “being completely kicked over.” “People that will join companies next year will be managers on day one,” of “an agentic-based workforce, one that is extremely powerful and sometimes clumsy.” Work runs human, then agent, then human, and delegation becomes a skill to train.
  5. EY’s hardest change is the billable hour. EY invested more than $1 billion in a year, and leadership buy-in worked. The hard part is a business model where “the harder the problem, the harder you have to work, the more you get paid,” and its clients have to change along with it.

Key claims

A 100-year firm, and the consultant as creator

EY has 400,000 people and “100 years’ worth of history.” A couple of months after ChatGPT appeared, it concluded that AI “was going to impact every single corner” of the business and created EY.ai, “the symbol for us internally that we need to move from the old business EY.com to EY.ai.” Diasio’s job is four things: moving client work “from this idea of ideation and experimentation into enterprise scale”; infusing AI into EY’s own delivery, since professional services sit “near the top” of the industries AI will hit; commercial model innovation, breaking “a business model that is focused on effort”; and training, tooling and enablement for staff.

On the threat to knowledge work, he describes an initial “shock and awe,” then a settled view: “the knowledge worker is just totally underrated in this moment.” Consultants carry context from outside the client’s four walls, and that context is what AI needs. Their role shifts “from one that does a lot of work and produces content to one that is much more of a creator, and using agents to be able to do a lot of that work.”

The masterclass, and statistical sameness

Shortly after ChatGPT, EY saw clients “being bombarded by PowerPoints explaining what AI was.” It built a master class instead: EY staff came with laptops, simple tools and a script, often to the client’s boardroom. CEOs worked in teams with their management committees to build a new snack product, a marketing plan and a jingle, then pitched it. Across hundreds of sessions the outputs converged on the same packaging, ingredients and recipes: “AI makes it very easy to produce something good, but it kind of commoditizes the output… it raises the floor, but it doesn’t raise the ceiling.”

The remedy is context. EY’s “big four” are creativity, critical thinking, systems thinking and deep domain expertise. The same applies to agents: they “are still subject to the same statistical sameness if you use them to do your critical thinking,” and “the AI is going to create the same predictable output for you as it would for somebody else if you’re not using that context with it.”

Mindset, skill set, tool set

“The last couple of years have been an investment in technology.” Firms then ask whether the tool does what the person did before, “and then you get into this productivity loop. And unfortunately productivity has a limit… Growth has no limit.” Transformation “begins with the intent of how we start to frame the problem.” On where to start, he says to look at “where the company creates value”: product innovation and R&D for consumer-goods firms, something different for banks or oil and gas. Firms that test AI in the back office “get into this like cost out productivity,” while strategic companies take on “challenging business questions that are existential to their business strategy.”

The strategy page and the blueprint

The first artefact is “a point of view that you would be comfortable taking to your investors,” usually “2 to 3 pages of like just plain text.” A portfolio of use cases is gym work before the New York Marathon: “they’re helpful, but they’re not the marathon.” His example of such a point of view is reinventing marketing to target “people’s agents that are going to buy the products on their behalf.” Wood names two common responses: spread AI over everything, or run a few pilots and wait for others. Diasio says “the first two is what we’ve seen over the course of the last three years,” because firms are still learning the controls needed at scale.

EY’s blueprint for a function built “as if you’ve built it for the era in which we’re in today” has these layers:

  • connectivity to existing systems, rather than ripping out past technology investment;
  • an AI-native platform;
  • an intelligence layer;
  • a control layer, “responsible for maintaining how much autonomy do you give these agents,” which also settles “where do you put the agents? Do you put the agents in the HR master, or do you put them in some other registration? And what access do they have?”;
  • then the process, “what the new jobs are of the future,” and where value is created.

Doing this in one area after another “allows you to work towards becoming the Frontier Firm.”

The competitors clients fear

“Most of my clients are really concerned about the competitors and the disruptors that they do not see… It’s the AI natives,” companies of “40 or 50 people” with “multibillion dollar valuations.” He describes ownership of AI moving over the preceding six months. It began as executive education, became the job of the CIO or CTO to scale, and is now “going back up to the CEO and the executive leaders… to set the vision and the strategy.”

Asked what an incumbent brings, he lists four things:

  • scale;
  • tolerance for “the messiness, which is a corporate reality” that AI-native start-ups dislike;
  • a data flywheel, because “organizational nuance and context is what drives differentiation from just using a general-purpose model”;
  • customer reach and contracts.

His example is an unnamed technology conglomerate whose most valuable AI asset turned out to be its game consoles and their community of players.

Upskilling, and a poll

Last year EY’s consulting arm built a curriculum for all staff: foundations first, then AI “in the context of their jobs.” EY calls the aim “the power of the and”: teaching people “how to find their ‘and’”, turning their domain expertise into a differentiator. He reports that employee engagement scores rose and that people “feel more in the driver’s seat… as opposed to have AI happen to them.” The investment is to be repeated every year.

He cites a poll: about 85% of employees “were really excited for agentic AI,” but 86% “had to learn it on their own… This is not EY specific. This is broadly across workers.” He does not say who was polled or when. His prescription has three parts: a strategy with clear intent, closer to growth than productivity; investment in skills and tooling; and change agility, because “this is going to be an ever-going cycle of change.”

Inside EY: what worked, and what is hard

Client Zero is EY’s programme for “doing this to ourselves first and then showing our clients what’s worked and what hasn’t.” What worked was leadership. The CEO is “very active in driving and shaping our AI agenda”, a direct report of hers leads the programmes across service lines, and priorities cascade to every geography. “Last year, we’ve invested over a billion dollars of capital into getting ourselves ready to become that Frontier Firm.”

What is hard is the business model. “We are a business model that has been built on the billable hour, where people monetize their efforts.” AI will “start to break that business model, but it fundamentally changes how we price our services and how we value our services.” Clients must come along too, and the firm must break “the muscle memory of… being in a business for decades under a consistent business model.”

Managers on day one

“This is probably the biggest revolution of management that has ever happened in our society.” The old ladder ran from building credibility and experience to becoming a manager. Now new joiners will manage agents from the start, which needs the big four plus a fifth skill: “a real structured and a real investment in developing the delegation skills. Because giving agents instructions to be able to go do the work, and then critically evaluating if that happened in the appropriate way is critical.” Agents take “the toil that nobody wants to do,” but “the real sequence of work should be human, agent, and then human.”

He closes with the internet: “we were really worried that it was going to replace all the mail carriers. And then it created trillion-dollar businesses.” In every technological revolution there is “first fear and then reinvention. And I think we’re just coming out of that fear phase right now.”

Dynamic-capabilities reading

  • digital-sensing/digital-mindset-crafting: the mindset–skill set–tool set triad, CEO masterclasses that put management’s hands on keyboards, and the two-to-three-page point of view as the long-term digital vision.
  • digital-transforming/improving-digital-maturity: a curriculum for every employee that teaches AI in the context of each job, and the finding that 86% were learning on their own.
  • digital-transforming/redesigning-internal-structures: the blueprint’s layers, with a control layer that decides agents’ autonomy, registration and access, and the new jobs it implies.
  • strategic-renewal/business-model: commercial model innovation, breaking the billable hour, and repricing services by value rather than effort.
  • contextual/external-triggers: AI-native disruptors of 40 to 50 people with multibillion-dollar valuations, and professional services near the top of the industries AI will change.

How it connects

  • enterprise-ai-adoption — mindset, skill set and tool set; from use cases to reinvention; the blueprint; Client Zero.
  • micro-productivity-trap — “productivity has a limit”, and back-office experiments that end in cost-cutting.
  • durable-skills — “AI and something”: the big four, delegation, and “the power of the and.”
  • agent-oversight-and-delegation — human, agent, human; a control layer for autonomy; delegation as a trained skill.
  • ai-employment-effects — new joiners as managers of agents, and the knowledge worker as underrated.
  • strategy — a strategy written for investors, and what incumbents hold against AI natives.
  • Sternfels (McKinsey) — McKinsey’s managing partner on a workforce of 40,000 people and 20,000 agents, and on a third of revenue already priced on outcomes.
  • Warren (YC) — AI-native services companies in tax, audit, law and insurance, which sell the outcome and price on it rather than by the hour.
  • The Frontier Firm AI Initiative’s About page — the initiative EY belongs to. Its early insights say employees will not find uses on their own without guidance, and that agents add a need for “agent management skills.”
  • Bornet, Miller, Sadun and George — other WorkLab episodes with Molly Wood on the Frontier Firm. Bornet built intelligent-automation practices at EY and McKinsey.
  • Mollick (HBR IdeaCast) — automation by default and “the same [Claude]-based company everyone else does”; “AI skills look like management skills.”
  • Kropp et al. (BCG, HBR) — a randomised experiment on what calling AI an employee or teammate does to managers’ accountability and error-checking. Diasio speaks of an “agentic-based workforce” and asks whether agents belong in the HR master.
  • Weidmann, Xu & Deming — a lab measurement of how well people lead teams of AI agents, compared with teams of humans.
  • Giles (WP Intelligence) — an EY-sponsored report whose ads promote “EY.ai Value Blueprints.”

What was actually ingested

The full episode (29:55, length_seconds: 1795) from the human-written caption track, 1,058 lines. One caption error was fixed in the raw file: “have I happen to them” is “have AI happen to them.” The episode is from December 2025, ten months before ingest, so “next year” means 2026 and “last year” means 2024 or 2025. No stills: it is a podcast.

Linked entities and concepts

Dangling (single-source mention, deferred): Dan Diasio, Molly Wood (host, never author:).

Scope and reliability

An EY executive describes EY’s own transformation, and EY’s consulting approach, on the podcast of a company that co-runs the initiative EY belongs to. The figures (400,000 people, more than $1 billion in a year, the 85% and 86% poll) are EY’s own and unsourced in the episode. The poll’s population is not stated. The sameness of the masterclass outputs is an observation across “hundreds” of sessions, not a measured result. “Managers on day one” is a prediction for 2026. The episode is useful because an incumbent consultancy names its own pricing model as the hardest thing to change, and because it gives a specific shape to the blueprint: a control layer that owns agent autonomy and registration.

Relationships

Supports

Graph View

Backlinks