Talent study · 10/3/2026
Seven connected findings from 12,142 job postings at 203 employers across 37 countries. This is a snapshot of advertised openings, not a census of the workforce.
The signal worth testing
AI has entered the hiring pitch. Hands-on demand concentrates in builders, Product and senior specialists.
The useful question is no longer whether a posting mentions AI, but where it asks for something concrete.
Chapter 01 · AI language
46% of postings mention AI or automation, yet only 10% name a tool, model or method. The stronger signal is where those concrete requirements cluster.
mention AI or automation
reference AI skills in a broad way
name a AI tool, model or method
AI language has spread across hiring, while specific implementation requirements remain concentrated.
AI postings as a share of each role family.
AI postings as a share of each seniority band.
AI postings as a share of each classified sector.
Infrastructure/ tools names in the 1,218 AI openings.
The broader picture: AI demand concentrates in Technology, Product, and Senior Specialist roles. Broad AI language has spread much further than hands-on requirements.
Chapter 02 · Job pools
We reviewed 12,142 current job postings in this quarter's collection. That shows today's advertised mix. A separate U.S. employment series shows how specific occupation pools moved from 2016 to 2024.
Current quarter · advertised openings
Every posting currently captured for this study edition is counted below. This is a snapshot of demand this quarter, not a workforce census or a trend.
Share of 12,142 job postings in this collection. Every classified family is shown.
Shares of business / functional postings only; ‘Other business’ remains unclassified.
Current quarter · openings by sector
Client-facing roles are 31% of collected wealth-management openings versus 7% in asset management. That fits their different customer models, but these postings show association, not cause.
Shares within this quarter's collected openings. They are not workforce shares or evidence that a sector is growing.
Change over time · employed people
This does not show that “operations” or “client-facing work” moved in one direction. Individual pools moved differently, even inside the same broad family.
U.S. employment by role · 2016 → 2024
Working interpretation: Processing work thinned while advice, analysis and technology grew. The sector added staff overall, so this is a change in mix, not a collapse of operations. Selected occupations cannot tell us whether the whole operations function grew or shrank.
Source: BLS Occupational Employment and Wage Statistics ↗, national industry estimates, May 2016 and May 2024. The May 2025 estimates were released in May 2026 and will replace the 2024 column once verified row by row. Since 2022, BLS publishes wealth, asset management and brokerage only as one combined industry, so they cannot be separated here. Banking combines depository banks and related activities. These are employed people, not job openings, and they don't show causes.
The broader picture: The evidence supports a shift away from some routine processing pools towards advice, analysis, control and technology. It does not support saying that all operations or all client-facing work declined.
Chapter 03 · Technology & Product
Product management is 2.7% of this quarter's openings. Product owners, analysts and delivery roles are counted separately. Each group's direction comes from outside research, because one snapshot cannot show a trend.
Technology · 2,654 openings (role count)
Product & delivery
of ask for Agile or Scrum
Product work is getting leaner, not disappearing. AI now drafts the stories, specs and mockups, so the process layer thins first. Someone still has to validate the spec, talk to users and own the domain. Today 2.7% of openings are product management alone and 5.2% with owners, analysts and delivery roles. We expect the generalist share to fall. That expectation comes from outside research; this snapshot cannot measure it.
Product & delivery, split by what the title owns
Product management
Owns: What gets built and why
TREND: LEANER
Open PM roles recovered in 2025 to 54% above the 2023 low (TrueUp ↗), but hiring tilts senior, and AI-first companies post about a third fewer PM titles per opening — a single-day sample (Riso Group ↗).
Product ownership
Owns: A team's backlog
TREND: MIXED
No reliable trend series. Backlog writing and story drafting are among PMs' biggest AI time savings (Productboard, 379 PMs ↗), which points to fewer, broader owners — an expectation, not a measured decline.
Business & systems analysis
Owns: Requirements and process detail
TREND: UNKNOWN
No finance-specific trend data found. Requirements work is exposed to the same AI drafting, but validation with users and controls stays human.
Agile & delivery management
Owns: How the work flows
TREND: SHRINKING
Scrum masters were among the fastest-shrinking tech roles in 2024 (TrueUp / Live Data ↗); Capital One cut 1,100 Agile-family roles in 2023 (Banking Dive ↗).
Counts are this quarter's openings; the trend label is outside research only. Business and systems analysts sit inside technology teams and own requirements, not the product, so they are reported next to Product, not inside it. Designers (70), product marketing (27), product control and production support are excluded.
Agile in the data
Agile or Scrum shows up in 10.9% of all openings, and most of those mentions sit inside technology ads (693 of them) — a baseline expectation of engineering work, not a product credential. Meanwhile dedicated Agile titles are nearly extinct here: 15 of 12,142 openings carry one. The method survived; the job title did not — matching the wider market, where scrum-master roles were among the fastest-shrinking tech roles in 2024 (TrueUp / Live Data ↗) and Capital One cut its Agile job family in 2023 (Banking Dive ↗).
Supported
Every PM surveyed uses AI; the biggest time savings are PRDs, decks, research and stories. Freed time moves to decisions — the generalist "process" layer thins first. Productboard ↗
Mixed
Open Product roles recovered from the 2023 trough in 2025. “Product is disappearing” is too strong; “Product is getting leaner and more senior” fits. TrueUp analysis ↗
Supported, with limits
Carta put average U.S. startup Product new-hire salary at $189,000 in June 2025, tied with engineering. Specs still need validating and users still need talking to. Not finance-specific. Carta ↗
No finance-specific source yet shows domain PMs growing while generalists shrink. “Only domain experts remain” is a hypothesis. Of today's product-management titles, 94 name a financial domain and 24 are bare generic titles.
Who owns the build?
Citi’s stated target for reducing contractors as a share of its technology workforce, with no precise deadline disclosed. Reuters ↗
Citizens is building a 1,000-person capability centre with Cognizant. Captive teams and service providers can expand together. Citizens/Cognizant ↗
In a 2025 UK and Ireland survey, 26% of financial-services respondents planned to shift external-provider spend toward insourcing — a meaningful minority, not the whole market. Whitelane ↗
The defensible read: large institutions want more control of architecture, data and critical delivery. They still buy implementation, managed services and specialist capacity. This study did not collect TCS, Cognizant, Accenture, Infosys, Wipro or HCL postings, so this sourcing section comes from outside research.
The broader picture: Product has not disappeared. The defensible shift is away from a separate process layer and towards fewer roles carrying clearer technical, platform or financial-domain responsibility.
Chapter 04 · Employers
The useful comparison is not which keywords appear, but which kinds of work each employer type is hiring for, and how specific their technology asks are. All figures use the employer-balanced sample.
Tech: Fintechs 30% vs Banks 16% Business/ Functional: Fintech 30% vs Banks 43%
Banks & incumbents · 2,692 postings · 95 employers
Fintechs & others · 2,633 postings · 108 employers
Why it matters: the employer type decides which door is widest. Moving between the two is easiest in the families both hire for in similar proportions.
Employer-balanced sample: at most 35 openings per employer, so a few large banks cannot set the picture.
Share of each group’s job postings in this study. Choose a comparison to see the employers behind it.
Banks & incumbents · 2,692 postings · 95 employers
Fintechs & others · 2,633 postings · 108 employers
Fintechs · 2.1% percentage-point lead
Why it matters: cloud and data-platform tools are the skills most likely to move between banks and fintechs.
Banks: LSEG (16) · CME Group (14) · State Street (13)
Fintechs: Altruist (11) · Schonfeld (11) · Gemini (10)
Fintechs · 6.0% percentage-point lead
Why it matters: naming a model, tool or method signals budgeted AI work, not just ambition in the job ad.
Banks: BNY (9) · Vanguard (8) · Lombard Odier (8)
Fintechs: Gemini (34) · Juniper Square (22) · ComplyAdvantage (20)
Fintechs · 15.3% percentage-point lead
Why it matters: broad AI wording without specifics is closer to employer branding than a skills requirement.
Banks: CIBC (34) · Broadridge (33) · BlackRock (31)
Fintechs: Wealth Enhancement Group (35) · SEI (34) · Neo Financial (33)
Fintechs · 3.3% percentage-point lead
Why it matters: a higher senior share means fewer entry points for career switchers.
Banks: BNY (29) · DTCC (28) · State Street (27)
Fintechs: Nubank (31) · iCapital (29) · Trading 212 (27)
Shares are within each employer group, not a measure of headcount, spending or pay. Financial plumbing counts use confirmed financial-context mentions; ambiguous uses of “Swift” are excluded.
The broader picture: Employer type changes the role mix and how concretely AI is described. Fully remote work is the sharpest difference of all — see the note below. These are differences in advertised openings, not proof that either group uses more technology in practice.
Chapter 05 · Protocols
After checking original posting text, 38 postings clearly refer to SWIFT and 36 name ACH. The ranking below concerns verified mentions, not how often firms actually use each network. The broader cloud/platform measure (30% of technology and operations postings) uses a different definition.
SWIFT leads this sample — narrowly.
38 confirmed mentions, versus 36 for ACH. That is 28% of 138 named protocol mentions, not a share of the whole industry.
113 of 4,392 technology and operations postings name at least one of these requirements, across 61 employers. A posting can name more than one; percentages below are shares of mentions and may overlap at posting level. “SWIFT” is counted only with financial context in the original text.
Most common in: Operations (32) · Technology (6)
Also ask for cloud: FIS (1) · State Street (1) · TD Bank (1)
Most common in: Operations (25) · Technology (11)
Also ask for cloud: Bank of America (1) · MUFG (1) · Gemini (1)
Most common in: Technology (19) · Operations (7)
Also ask for cloud: State Street (3) · LSEG (2) · FIS (1)
Most common in: Operations (10) · Technology (9)
Also ask for cloud: FIS (2) · Bank of America (1) · Alpaca (1)
Most common in: Technology (11) · Operations (3)
Also ask for cloud: TD Bank (2) · Wellington Management (1) · State Street (1)
Most common in: Operations (4)
Also ask for cloud: No named employer pairing in this sample
Most common in: Operations (1)
Also ask for cloud: No named employer pairing in this sample
Broader, unverified text mentions; not the validated financial-protocol ranking above. A posting can name more than one.
The broader picture: A protocol can be essential without appearing in many job descriptions. Absence from a posting is not evidence of absence from a firm's systems.
Move from the systems named to the cities where these openings appear.
Chapter 06 · Cities
Across Pune, Bangalore, Hyderabad, Mumbai, 56% of 620 located postings are technology or operations and just 1% face clients. In San Francisco, 79% of 335 postings come from fintechs, spread across technology (30%), business roles (34%) and client-facing work (16%). Operations is not offshore-only: New York 15%, London 16%, San Francisco 11% of postings are operations roles, against 19% in the Indian hubs.
Postings located in each city in this collection; not a ranking of local employment.
New York is a fintech company town — 51% of job postings here come from fintech employers.
Pick two hubs and compare what kind of work happens there.
The broader picture: Bank-owned capability centres in India are the banks' own staff, concentrated in technology, operations and control work, and they run alongside onshore teams rather than processing everything for them. City patterns reflect the employers covered, not a city's whole economy. Explore the employer map →
Chapter 07 · Technology tools
These are tool mentions in the collected job descriptions, not proof of proficiency or a ranking of what pays most. Each count is the number of postings naming the tool.
Share of 12,142 sampled openings naming each tool; postings may name several.
The broader picture: The technology list is a snapshot of stated requirements, not a forecast of which skill to learn next. Explore the skills map →
Separate reference · qualifications
29% of client-facing postings name at least one licence or credential. The list below counts mentions across all job postings; it mixes regulated licences and professional certifications, so it is not a measure of who must hold a licence.
Mentions across the sample; a posting may name more than one.
Eight claims, one reality check
Do banks really move most tech hiring offshore? Is finance plumbing requested more often than cloud? The measured answer sits under each claim; open it for the sample and limits.
The sample met the test's stated threshold.
49% mention AI; only 23% of those name something concrete.
Test sample: 5,323 job postings. Thresholds are stated where applicable. This tests collected openings, not the whole workforce.
Credentials named: client roles 26% vs tech roles 2%.
Test sample: 2,244 job postings. Thresholds are stated where applicable. This tests collected openings, not the whole workforce.
The observed share missed the threshold.
22% of incumbent tech postings list an offshore hub (needs 50%+).
Test sample: 439 job postings. Thresholds are stated where applicable. This tests collected openings, not the whole workforce.
Ops + client roles: fintechs 32% vs incumbents 36%.
Test sample: 5,323 job postings. Thresholds are stated where applicable. This tests collected openings, not the whole workforce.
Tech share in universal banks: wealth 2% vs banking 14% (within 8 points = similar).
Test sample: 737 job postings. Thresholds are stated where applicable. This tests collected openings, not the whole workforce.
8% name a standard or file format vs 28% name a cloud platform.
Test sample: 2,037 job postings. Thresholds are stated where applicable. This tests collected openings, not the whole workforce.
1% name both a legacy and a modern platform (needs 5%+).
Test sample: 5,323 job postings. Thresholds are stated where applicable. This tests collected openings, not the whole workforce.
The sample is too thin for this test.
Advertised annual base-pay range midpoints (USD): hybrid insufficient sample (n=17) vs pure tech $180k (n=130). Not total compensation.
Test sample: 147 job postings. Thresholds are stated where applicable. This tests collected openings, not the whole workforce.
Now test yourself
Quick quiz
Three questions. Guess first, then see what the job postings say.
1. Which is rarer in financial-services hiring?
2. Do fintechs give a much smaller share of hiring to operations and client support than banks?
3. In tech and operations postings, which is named more often?
Postings listing several countries count in each country.
