Share with your CIO
Vanguard is betting that AI at enterprise scale in financial services lives or dies on foundational discipline, not model sophistication. Jing Wang, who leads Vanguard’s Center for AI and Insights, made that case at MIT Technology Review’s EmTech Future conference, arguing that data, engineering, platforms, operating models, and governance must function as a single integrated system before agentic AI, meaning AI that acts autonomously on multi-step tasks, can deliver durable value. The session is framed around regulated-industry constraints and investor outcomes.
What this means for your business
Based on limited public information from MIT Technology Review, the substance of Wang’s argument is thinner on the page than the framing promises. What comes through clearly is a conviction most regulated-industry CIOs already hold but rarely act on fast enough: the organizations that will actually run agentic systems at scale in the next two to three years are the ones that treated governance and data infrastructure as prerequisites, not retrofits. If your AI roadmap still treats governance as a compliance checkbox appended after the build, this is a signal you’re on the wrong side of that split.
The Vanguard framing, coming from a firm whose entire brand proposition is low-cost fiduciary trust, naturally emphasizes responsible and trustworthy AI over speed-to-deploy. That’s not a distortion, it’s a real constraint, but it does mean the playbook Wang describes is calibrated for organizations where regulatory exposure is existential. For CIOs in less tightly regulated sectors, the risk calculus shifts: moving slower than Vanguard on governance is acceptable, but borrowing their foundation-first sequencing still holds, because agentic systems that act autonomously on client data without clean lineage, meaning a clear record of where data came from and how it was transformed, create audit nightmares that compound fast.
The leading indicator to watch is whether your current data and platform teams can answer, right now, which systems an agent would touch, in what order, and under what authorization rules. Organizations that can answer that question have already done the foundational work. Organizations that can’t aren’t one model upgrade away from readiness; they’re one incident away from a mandated pause. That’s the budget decision this reframes: not whether to fund the next AI initiative, but whether the infrastructure underneath it can bear the weight of autonomous action.
Based on reporting from The Foundations of Scalable AI, originally published 2026-10-22 14:48:00.
