Share with your CEO
EY is betting that India, where more than 100,000 of its 400,000 employees sit, will be the operational core of its global AI build-out. CEO Janet Truncale launched the EY.ai Centre for Reimagination in Bengaluru, a client-facing experience centre backed by the firm’s $1.4 billion AI investment. The centre’s explicit purpose is to pull enterprise clients past the pilot trap, the pattern where companies accumulate narrow AI wins but never string them into business-model change. EY is also deploying agentic AI across its own audit platform, Canvas, which serves 160,000 clients.
What this means for your business
The pilot trap is the dominant failure mode in enterprise AI right now, and Truncale’s candid summary of client complaints lands accurately. CEOs are telling her that two years of use cases produced productivity gains but not margin relief, because technology costs rose in parallel. If your organization is deep in that pattern, the meaningful question isn’t whether to add more pilots. It’s whether your operating model can absorb end-to-end redesign, which requires a different kind of internal sponsorship than a proof-of-concept budget.
EY’s “client zero” framing deserves scrutiny. Using your own audit and tax workflows as the test bed before selling the same approach to clients is genuinely good epistemics, not just marketing. The Canvas agentic layer matters here because audit is one of the most process-dense, liability-heavy workflows in professional services. If agentic AI, meaning software agents that execute multi-step tasks autonomously rather than just answering questions, can run end-to-end in that environment without audit failures, the credibility transfer to client engagements is real. EY has an obvious incentive to present this progress optimistically, but the concrete platform name and the 100,000-user scale give it more weight than a generic transformation claim.
The talent argument is where EY is most exposed to being wrong, and most interesting to watch. Truncale says junior hiring at EY’s Global Delivery Services ran to 25,000 people in the past year alone, and that AI elevates rather than eliminates entry-level roles. That may hold while billable work expands faster than AI compresses it, but the agentification of country-by-country tax knowledge catalogues is exactly the kind of work that used to require armies of junior analysts. The falsification condition is straightforward: if GDS campus hiring drops two years from now while revenue per head rises sharply, the “elevation not elimination” thesis will have failed its own test.
Concept deep-dive: Agentic AI
Agentic AI refers to systems that don’t just respond to a single prompt but execute sequences of decisions and actions autonomously, closer to an employee following a workflow than a search engine answering a query. It exists because large language models alone are stateless and reactive, while real business processes require memory, tool use, and multi-step judgment. On EY’s Canvas platform, an agentic layer means an auditor can ask a question and have the system pull relevant standards, cross-check prior filings, and surface a recommendation in one pass.
Based on reporting from EY CEO Janet Truncale says India is central to the professional services firm’s AI strategy for the world, originally published 2026-06-11 20:21:00.
