AI gets an org-chart slot at Wells Fargo

A bank gives AI a reporting line

Wells Fargo announced that Saul Van Beurden will focus exclusively on leading AI across the company, working with Bridget Engle, the head of technology, to drive enterprise-scale adoption. Van Beurden previously co-led Consumer Banking and Lending with Kleber Santos; Santos now becomes the unit's sole CEO, overseeing consumer deposits, credit cards, business banking and auto lending. Both report to chairman and CEO Charlie Scharf. The bank has about $2.3 trillion in assets. Van Beurden established and led Wells Fargo's AI efforts before this announcement. m:…1115

The use cases are already decided

If the org chart is new, the work isn't. WorkFusion's compliance team argues AI agents are now suited to specialized financial-crime work — name screening, transaction monitoring, know-your-customer checks — with false-positive management as the clearest opportunity. The pitch is scale without headcount: compliance teams face persistent backlogs from growing alert volumes and can't simply hire their way out. Governance, the firm says, is the most important consideration, and compliance professionals must remain accountable for decisions requiring human judgment. m:…1177

In FP&A, Pigment's Glenn Hopper offers a cleaner division of labor: code does the math, AI investigates and drafts, people make the call — his words. His framework for delegating work to agents weighs magnitude, frequency and recoverability, and insists on an 'agent harness': tools, guardrails, approval gates and a full evidence trail. The warning is pointed: a human approval gate only works if the reviewer can see the evidence and their decision flows into the forecast. A language model's draft, he notes, is fluent, confident and plausible — which is exactly what makes it dangerous when it's wrong. m:…8225

Governance catches up

FloQast's answer to the same anxiety is institutional: a new COSO AI module that lets companies record their AI agents, the internal controls they support, and the responsible parties — accounting's version of putting agents on the books. The firm says finance teams must lead adoption or be saddled with systems that create more problems than they solve, with accuracy, budget and compliance as the governing principles. Its advice: start with low-risk, high-visibility processes, and ask what the automation strategy is before what's the AI strategy. The same announcement includes a five-level AI maturity model in FloQast's "The State of Accounting AI 2026" report, plus an AI Assistant that checks journal entries for errors and assigns risk scores based on historical data. m:…0081

The demand signal

Goldman's 2026 alumni survey of its 10,000 Small Businesses program suggests the rewiring isn't just an enterprise story: 42% of respondents now use AI for financial analysis and business planning, more than double the 2024 rate, and over 85% use AI in some form. One example: McClure's Pickles cut ingredient-origin trace times from 30–45 minutes to under two minutes with LLMs. Goldman is launching a Small Business AI Institute to help owners navigate adoption. m:…4406

This suggests the pattern for an AI executive: the experimental phase is ending and the accountability phase is beginning. Banks are creating AI roles with enterprise-wide authority, vendors are selling governance as the product, and accountants are being told to own the controls. The winners won't be the firms with the most agents — they'll be the ones whose agents can survive an audit. m:…1115 m:…0081