AI in Finance — Workforce Digest – Week 28 July 2026

A single-carrier run. Three of the five items fall inside the 20–27 July window, and all three reach the log through one article — Investment News's wealth-technology feature of 24 July — which relays three separate primary sources: the F2 Strategy Q2 2026 Trend Report, the Crisil Coalition Greenwich “US Equity Electronic Trading: Broker View 2026,” and Bank of America's Erica Assist enhancement. Two further items, both dated 30 June, predate the window and are carried for the wealth and entry-level threads, with their dates noted inline. None of the five is a headcount-reduction announcement — a reversal from the prior run's executed Indian FY26 net cuts. Provenance notes indicate the strength of sourcing, not an endorsement of the claims.

AI spenders are hiring

US equity trading desks (Crisil Coalition Greenwich, “US Equity Electronic Trading: Broker View 2026,” July) — More than half of the US brokers surveyed expect to increase headcount even as AI adoption across trading workflows accelerates: desk coverage 52%, on-desk trade assistants 48%, algorithmic sales 45%. AI use is rising in parallel — real-time algorithm optimisation 32%, venue selection 29%, market-data analysis 29% — with compliance and surveillance at 12% today and 44% planning it. The report ties the planned hiring to near-record US equity volumes and the IPO pipeline (SpaceX and OpenAI named), positioning AI as concurrent adoption rather than the driver, and states that “the human element is actually becoming more central, not less,” with value shifting to “judgment, client relationships and exception management” while automation “handles routine tasks.”

Confirmed as the firm’s own report; the figures are reported via InvestmentNews and refer to participating brokers, not the sector. The hiring is attributed by the source itself to volumes and the IPO pipeline, not to AI. Read this run only through the InvestmentNews feature; the standalone report is flagged for direct capture.

The biggest AI spenders (American Banker, 30 June) — Across 21,559 US firms studied by Ramp and Revelio Labs, those spending most on AI per employee saw headcount rise 10.2% in the two years after adoption — entry-level up 12%, non-entry-level up 7.7%, managers-and-above up 6.7% — while firms spending little or nothing saw no increase. PwC's 2026 AI Jobs Barometer, cited in the same piece, finds banks hiring AI expert users over model creators (88% versus 12% of AI-related financial-services listings in 2025), AI mentions in FS postings up from 3.9% to 6.1%, and a 54% wage premium. Lloyds Banking Group said in late June it will recruit for almost 300 agentic-AI roles on top of more than 700 employees already doing the work.

The article is confirmed as published; the Ramp/Revelio and PwC findings are estimate/model-grade — correlational, and unable to separate AI adoption from firms’ underlying growth. The Ramp/Revelio dataset is the same one already logged on 14 July, where it was cited to show operations was the only job category not growing headcount among heavy AI adopters; this article relays the aggregate “hiring more” reading and does not reconcile it with that operations exception. The Lloyds ~300-role recruitment is confirmed as the bank’s own statement. Dated 30 June; captured this run.

A counter-line inside the same piece (Jeff McMillan) — Within that American Banker article, the former Morgan Stanley firm-wide-AI head Jeff McMillan says current copilots deliver roughly 5% efficiency — not the ~48% employees self-report — and predicts the next recession will bring “the largest percentage of layoffs that we've ever had,” with those jobs potentially not returning.

Reported: a named executive’s public characterisation and prediction, not an independent measurement. It sits in the same article as the “hiring more” data, on a different time horizon; the two are not reconciled.

Augmentation deployed, headcount trajectory unstated

Bank of America, EricaAssist (announced week of 20 July) — Bank of America enhanced EricaAssist, its generative-AI tool supporting more than 18,000 customer-service representatives, to deliver contextual guidance in under three seconds during client calls. The bank says it already reduces average call times by nearly one minute per interaction and will expand it to more servicing scenarios and business lines later in 2026. Bank of America states it spends $14bn a year on technology, more than $4bn of it on new initiatives including AI, and frames the tool as “high-tech, high-touch” — augmenting representatives rather than reducing them.

Confirmed as Bank of America’s own newsroom announcement; the per-call saving and the spend figures are as stated. No headcount figure is attached, and the disclosure does not say whether the per-call saving will change hiring for the 18,000-rep tier it augments — the trajectory is unstated, not disputed. This is the same firm’s Erica line last logged as the internal “Erica for Employees,” now customer-facing; it joins the Citi Sky and Barclays deployments as augmentation announced with no staffing follow-through.

Wealth management: the back-office and paraplanner tier

RIA hiring reshaped (Investment News, “AI is reshaping wealth management hiring, but human advice remains safe,” 30 June) — Three named RIA leaders describe AI compressing administrative, back-office and marketing/content headcount while adviser, planner and relationship roles “remain beyond automation for now.” Andrew Mescon (Ballast Rock Private Wealth) has added AI fluency to hiring criteria, says he will not cut back-office staff but is less likely to add marketing headcount, and flags AI’s potential to erode financial-planning workflows and reduce demand for paraplanners. Ethan Jung (Abundo Wealth) built proprietary in-house tools rather than hiring for those functions. Cited within the piece: a 2024 SHRM study (78% of hiring managers said AI use will lead to layoffs of recent graduates), a 2026 Oliver Wyman analysis (a bifurcating market where white-glove human service commands a premium), and a Wealth Management magazine figure (one support hire serviced 86 clients / $517,500 revenue in 2022 versus 111 clients / $591,000 by 2024 with the same team, attributed to technology).

Reported: three RIA-leader interviews plus cited third-party studies. No back-office or paraplanner headcount reduction is quantified and no horizon is given. This is the only source in the run to name the exposed tier explicitly — administrative, back-office, marketing, paraplanner — and it sizes none of it; the paraplanner erosion is named as a watch-item. Dated 30 June; captured this run.

Spend up, ROI unmeasured

F2 Strategy Q2 2026 Trend Report (via Investment News, 24 July) — AI spending across wealth firms has risen “exponentially” over three years while most have no formal method of measuring AI ROI — and none of the bank-and-trust respondents had one. The report finds 64% of wealth-management firms and 83% of bank-and-trust respondents lack a unified data layer to make AI projects function; among the firms that do measure, 68% report “25% more efficiency in targeted workflows”; and it draws a widening divide between “AI leaders” building agentic stacks and laggards 12–24 months behind. The workforce enters the report only as that efficiency figure, with no headcount attached.

Confirmed as F2 Strategy’s own Q2 2026 Trend Report; the efficiency figures are survey self-reports (estimate/model). Alongside McMillan’s ~5% measured copilot efficiency above, it is the second efficiency-scepticism datum in the run: budgets surge while ROI is either unmeasured or, where measured externally, smaller than self-reports claim. Read this run only through the InvestmentNews feature; the standalone report is flagged for direct capture.

Still open

No executed-headcount data this run — a reversal from the prior week's HDFC/Axis/Kotak FY26 net figures; the Q2 2026 US-bank 10-Qs with disaggregated headcount were still not available in open sources. No new named-firm AI-attributed headcount number: Dimon's “30–40% in certain areas” resurfaced via The Motley Fool on 21 July, unchanged. No new UK or US regulator output — the 20-July UK instruments (HM Treasury's Financial Services AI Adoption Plan and the Skills Compact) gained no new signatory or implementation news this window, and there was no new FCA, PRA, Bank of England, HM Treasury, SEC, US Treasury or Federal Reserve workforce output, the eighth consecutive run without US regulator output. Still undeveloped: HSBC's roughly 20,000 roles under consideration (about ten weeks without development); the BlackRock ~250-versus-~500 reconciliation (unresolved since 15 June); Standard Chartered beyond its restated figure; the Citi Sky and Barclays deployments, now joined by EricaAssist as a third augmentation with no headcount trajectory; the Bloomberg mass-affluent service-threshold question; and private-equity junior compression (unverified since the baseline). Flagged for direct capture next run: the F2 Strategy Q2 2026 Trend Report and the Crisil Coalition Greenwich Broker View 2026 as standalone primary documents, and the Morgan Stanley “Echoes Past Tech Transitions” note.

Assembled from a structured weekly monitoring run. Items reflect what was reported during the week; provenance notes indicate the strength of sourcing, not an endorsement of the claims.

SMA Crown Confidential

Digital Confidantes: Bespoke AI Intelligence for Private Decision-Makers