The wave of AI pilot projects in the wealth-management industry is not new. What is new is who names the success on which stage. On 26 March 2026, Bank of America (BofA) announced the firm-wide roll-out of an Agentic AI tool for its wealth-management businesses Merrill Wealth Management and Bank of America Private Bank. Three weeks later, on 16 April, Chief Financial Officer (CFO) Alastair Borthwick – on a media call, not on the official earnings call held the day before – positioned the tool by name as a driver of the quarter's results. With that, BofA becomes the first US Tier-1 wealth-management player to couple an AI investment explicitly, by name and through the CFO, to the earnings narrative. It is not the most spectacular scaling – but it is the most consequential narrative.
Tool: AI-Powered Meeting Journey – preparation, live note-taking and follow-up for advisor-client meetings
Scaling: Beta with more than 1,500 advisors, full-scale roll-out from 26 March 2026 to roughly 18,000 advisors at Merrill Wealth Management and Bank of America Private Bank, completed by mid-April 2026
Tech basis: The BofA press release names Salesforce CRM and Salesforce Financial Services Cloud; the “Agentforce” branding stems from third-party reporting, not from the bank's own communication
Earnings link: Q1 2026 net income of $8.6 billion (up 17 per cent year-on-year), Earnings per Share (EPS) of $1.11 (up 25 per cent year-on-year – the divergence is explained by $7.2 billion in share buybacks during the quarter)
Technology budget 2026: $13.5 billion annually, of which $4 billion is dedicated to new initiatives such as AI
What actually happened
BofA's press release of 26 March 2026 calls the tool “AI-Powered Meeting Journey” and describes it as an integrated solution that supports advisors in meeting preparation, conduct and follow-up. The functionality ranges from consolidating client information ahead of the meeting and AI-supported live note-taking during virtual meetings – with the client's consent – to the automated generation of summaries and follow-up tasks. The bank quantifies the time saving at up to four hours per meeting.
The bank's own communication is strikingly restrained on the technology stack. Patricio Diaz, Chief Operating Officer (COO) of Merrill Wealth Management, frames it this way: “AI-Powered Meeting Journey represents a meaningful advancement in how the wealth management industry uses AI.” The press release names Salesforce CRM, Salesforce Financial Services Cloud and Zoom as technology partners – but not “Agentforce”, Salesforce's current agentic product. The attribution as an Agentforce deployment originates from coverage by Banking Dive and has been picked up by industry newsletters such as NASCUS and MLQ.ai. This difference is not a semantic footnote. It is relevant because “Agentic AI” and “classic CRM with AI layers” carry different regulatory and operational consequences.
Three weeks later comes the move that turns the story into an earnings story. On 15 April 2026 BofA publishes its Q1 numbers: $8.6 billion in net income (up 17 per cent year-on-year), $1.11 in Earnings per Share (up 25 per cent year-on-year). The divergence between net-income growth and EPS growth is explained by $7.2 billion in share buybacks during the quarter. Chief Executive Officer (CEO) Brian Moynihan calls the EPS figure “the highest level in almost two decades” on the earnings call. But the comment about the tool comes a day later, on 16 April, on a separate media call. Alastair Borthwick describes the Meeting Journey tool concretely and says: “Efforts like this translate into results.” Elsewhere he adds that the tool's output does not replace judgement – “not necessarily the judgment–that can be human”.
What sets BofA apart from Morgan Stanley
Anyone tracking the US wealth-management industry knows that Morgan Stanley had the pilot earlier and pushed adoption deeper. The “AI @ Morgan Stanley Assistant” tool, based on OpenAI's GPT-4 with a Retrieval-Augmented Generation (RAG) architecture, has been in production since 2024 and reaches an adoption rate of 98 per cent across roughly 16,000 advisors, according to Morgan Stanley. That is the highest published wealth-AI adoption figure globally. But it is primarily a knowledge-retrieval tool: advisors ask questions, the system searches a curated knowledge base, an answer follows. It is not an agentic workflow.
Bank of America does something different – on two dimensions. First: the Meeting Journey tool covers a workflow, not just a retrieval. It consolidates data from multiple systems, acts as an AI note-taker, and generates follow-up tasks. Second, and more consequential for the industry: the CFO names the tool in the context of a record quarterly result. Morgan Stanley publishes AI adoption metrics, but does not couple them directly to its EPS argument. JPMorgan Chase operates the broadest enterprise AI platform in the industry through its LLM Suite – with around 200,000 firm-wide users on a multi-model basis (OpenAI plus Anthropic) and a technology budget of $19.8 billion in 2026. But the wealth-specific adoption and earnings figures are not disclosed. Goldman Sachs in February 2026 announced an Anthropic Claude partnership for trade accounting and Know Your Customer (KYC), without communicating a comparable wealth-management tool. Wells Fargo only launched its “Advisor Gateway” on 7 May 2026 – less than two weeks before this article – based on BlackRock Aladdin and a generative-AI auto-commentary feature. There is no earnings link there yet.
The Bank of America story is therefore not “Morgan Stanley but bigger”, but a different narrative about the same industry race. It shifts the discourse from “Who adopts AI in advisory most broadly?” to “Who manages to tie AI investments directly to quarterly results?”. That is a change of discipline that German wealth-management Chief Information Officers (CIOs) and Chief Operating Officers (COOs) cannot afford to ignore.
Salesforce, Agentforce and the branding question
In October 2024, Salesforce launched Agentforce, an orchestration platform for autonomous AI agents. The current version, Agentforce 360, was unveiled at the Dreamforce conference in 2025. The architecture is built on a proprietary Atlas Reasoning Engine, a ReAct loop implementation with YAML-declarative agent configuration. The default Large Language Model (LLM) is GPT-4o on Azure, with bring-your-own-model support for Anthropic Claude, Google Gemini, Amazon Nova and xAI Grok. For Q4 of fiscal year 2026, Salesforce reported Annual Recurring Revenue (ARR) of around $800 million for Agentforce, up 169 per cent year-on-year across 29,000 deals.
The nominal adopter lists look impressive at first glance. But on closer inspection, only one verified large-scale Tier-1 bank roll-out exists: ANZ in Australia (February 2026), in business banking rather than wealth management. Other customer stories – for example Royal Bank of Canada Wealth Management, Prudential Financial or Absa Bank – exist as Salesforce marketing material, but without independently verifiable scaling data. For JPMorgan Chase, Wells Fargo, Citigroup, HSBC or Standard Chartered, there are no documented Agentforce deployments in the wealth-management space.
Whether the BofA tool is therefore the first Tier-1 Agentforce roll-out in wealth management depends on how one handles the technology-stack attribution. The BofA press release itself does not say so. Banking Dive and the follow-on coverage do. Anyone who distinguishes between a Salesforce CRM layer and genuine Agentforce architecture arrives at a more sober verdict: BofA is the first to openly couple a Salesforce-based AI workflow for wealth advisors to its earnings narrative. The technology-firstness story is not provable. The earnings story is.
Three cracks in the success story
Anyone serving as CIO or COO of a German wealth-management house and trying to make sense of the BofA case should keep three issues distinct. First, the causality question. Q1 2026 EPS growth of 25 per cent is a result whose primary drivers are the trading business, Net Interest Income (NII) and the $7.2 billion in share buybacks already mentioned. AI has contributed to productivity, but Alastair Borthwick's statement – “efforts like this translate into results” – is a narrative coupling, not an analytical isolation. For internal investment cases, this is an important distinction: BofA is selling a correlation, not a causal model.
Second, adoption transparency. Morgan Stanley publishes an adoption rate of 98 per cent. BofA publishes 18,000 advisors as the target population, but no figure for daily usage frequency. For a tool that has only just gone into production, that is not unusual. For an earnings coupling, it is. Anyone wanting to copy the BofA narrative as a German wealth-management bank must build adoption measurement in from day one. Otherwise an asymmetry emerges: an earnings argument without an adoption proof is exposed to critical analyst follow-ups in the next quarter.
Third, the regulatory latency that is currently picking up speed in the US debate. Merrill itself, in its updated Wrap Fee Program Brochure filed with the Securities and Exchange Commission (SEC) in January 2026, explicitly disclosed AI risks: “AI tools are highly complex and may be flawed, hallucinate, reflect biases included in the data on which such tools are trained, be of poor quality, or be otherwise harmful, which therefore requires supervision and oversight.” Adding: “The legal and regulatory environment relating to the use of AI tools is uncertain and rapidly evolving, and could require changes in our implementation of AI tools and increase compliance costs and the risk of non-compliance.” The Financial Industry Regulatory Authority (FINRA) named AI agents explicitly as a risk focus for the first time in its 2026 Annual Regulatory Oversight Report: “tools that act without human sign-off, stretch beyond their intended authority, are hard to audit, or mishandle sensitive data”. The SEC and FINRA have placed AI on their 2026 examination priorities. There is no acute enforcement yet, but the structural risk is set.
The European lag – and what DACH wealth-management should take from this
The comparison with European houses is sobering. UBS has onboarded around 46,000 employees onto its Eliza platform (90 per cent of the workforce) and operates UBS Red, a productive wealth advisor assistant on a Microsoft Azure stack. The Credit Suisse acquisition provides the synergy basis for further investment. But there is no direct earnings link in the BofA mould. Deutsche Bank announced an “AI Banking Butler” at its Investor Deep Dive in November 2025 – voice-enabled, agentic, for 18 million personal-banking clients – but this is roadmap, not live deployment. The budget runs to roughly €600 million for IT, operations and AI over the period 2025 to 2028. BNP Paribas is further along in investment-banking AI (a pitch portal with Mistral AI from December 2025), but in the wealth space it is still in the implementation phase. Julius Baer and Commerzbank are clearly trailing.
The lag is structurally six to eight quarters, and it has three causes. On the regulatory side, the Digital Operational Resilience Act (DORA) since January 2025, the EU AI Act with its high-risk classification for suitability use cases from 2 August 2026, and the MiFID II requirements for explainability of advisory decisions are simultaneously a drag and a protective layer. Operationally, technology debt counts – Deutsche Bank is currently consolidating from 15 to 2 core banking systems. Commercially, finally, the absolute technology budgets of European houses are smaller while compliance costs are comparable. That makes the Return-on-Investment threshold (ROI) for AI investments less favourable.
That is not an excuse for inaction. It is a diagnosis. Anyone wanting to follow the BofA move must understand the differences they have to overcome. The trick is not to copy the American narrative – it is to translate the European drag conditions into advantages where that is possible. Explainability is already on the European mandate sheet, where in the US it is still an examination topic. Anyone addressing this proactively is not selling a story in 2027 – they are selling a demonstrated capability.
Four levers for DACH wealth-management CIOs and COOs
Four concrete tasks emerge from the BofA case that can be tackled in the next 30 to 90 days. They are not new in wording. They are new in pressure.
Which of your firm's AI investments would qualify for the CFO to name on the next earnings call? If the answer is “none”, what is missing is a measurement logic, not a tool. Establish a three-dimensional outcome measurement for every productive AI use case: productivity gain (hours, preparation time), quality gain (error rate, classification accuracy) and risk balance (added audit capability, lower compliance burden). Only this measurement logic makes the AI investment CFO-ready. BofA quantifies “up to four hours per meeting” multiplied by millions of meetings per year – that is the argument anchor, not the technology stack.
The EU AI Act will fully regulate high-risk suitability use cases from 2 August 2026. For wealth advisor tools that influence or document client suitability, that means: explainability, audit trails and bias monitoring belong in the architecture decisions made now, not in 2027. Anyone reading the Merrill SEC disclosure sees the pattern: even BofA acknowledges that “supervision and oversight” are required. Build the supervisory function and the explainability layer into the use case before marketing positions the tool as “Agentic AI”. Otherwise a compliance issue turns into a licensing issue.
Morgan Stanley publishes 98 per cent. BofA publishes no adoption rate. For your house: define before the roll-out which adoption metric you will publish – daily active users, share of features in productive use, advisor Net Promoter Score (NPS). This definition determines what is communicable in the next quarterly result. Adoption theatre – meaning high licence numbers without genuine usage frequency – is the most dangerous pseudo-success metric. It holds for 18 months, then the first analyst asks.
An AI workflow tool for wealth advisors fails on advisor acceptance, not on model quality. Both BofA and Morgan Stanley invest substantially in change management, training, and the message that the tool frees up advisor time for “deeper client conversations” – rather than replacing the advisors. Alastair Borthwick frames it as “not necessarily the judgment–that can be human”. For DACH wealth houses with smaller advisor populations and a different generational mix, this message matters even more. Structure the roll-out programme so that the first 200 advisors become advocates, not mandatory users. Scaling then follows adoption, not the deadline.
Timeline: The next 12 months
Bank of America's AI-Powered Meeting Journey is not the largest roll-out and not the deepest adoption. It is the most explicitly coupled earnings story in the wealth-AI industry. For German wealth-management CIOs and COOs the lever is not in copying the tool, but in building the measurement logic, the regulatory pre-answers and the advisor buy-in that make the narrative credible in the first place. The bar is no longer “do we have a wealth-AI tool?”, but “can we quantify its outcome?”. Anyone without an answer to that question two quarters from now has argued themselves out of the race.
Christian Schablitzki
Strategy & Management Consultant · Agentic AI Expert for Financial Institutions
More than 20 years in investment banking and derivatives trading, followed by over 10 years as a consultant to financial institutions. Currently a Partner at Infosys Consulting in Germany. Certified in Google AI, Generative AI Leader (Google Cloud) and IBM RAG and Agentic AI.
LinkedIn profile →Read on – every two weeks in your inbox.
Capital-markets insights, regulatory updates and AI trends. Concise, well-grounded, free.
GDPR-compliant. Unsubscribe anytime.