Method
Where AI changes the work, and how we rate it
AI impact follows what today's models are actually good at. Voice and maths lead, documents are next, and work built on organisational knowledge comes last. Funding and new entrants are evidence, never the score.
of Wealth capabilities are high impact (31 of 64)
of Asset capabilities are high impact (14 of 26)
Documents are next. Work that runs on unwritten firm knowledge changes last.
The heat map
Wealth
Changing fastest
- Account Aggregation
- Advice Engagement
- AI Assistant
- AI Notetaking
Changing slowest
- Advisory Fee Billing
- Business Valuation
- Caregiving & Special Needs
- Client File Sharing
Asset
Changing fastest
- Alternative Data
- Client & Fund Reporting
- ESG & Sustainability Data
- Execution Management (EMS)
Changing slowest
- Corporate Actions
- Custody & Depositary
- Fund Accounting & NAV
- Order Management (OMS)
How we rate it: what the research says
Speech & conversation
Most matureVoice and conversation is where AI already does real work: meeting notes, call summaries, voice-to-voice and answering simple client questions. AI notetakers are the fastest-growing advisor tech; over 4 in 10 advisors use generative AI.
Typical capabilities: Advisor CRM & notes, client service, chat and contact centres
Maths & structured reasoning
MatureCalculations, optimisation, matching and reconciliation suit models well. Large firms with cash and tech teams build these low-risk uses themselves — Goldman Sachs built AI agents for trade-accounting reconciliation with Anthropic.
Typical capabilities: Rebalancing, optimisation, risk analytics, reconciliation, surveillance
Reading documents
Next waveReading statements, factsheets, filings and forms is the next battleground. SIFMA and Sia point to document processing, alternatives reconciliation and corporate actions.
Typical capabilities: Onboarding & KYC, document intake, corporate actions
Writing documents
Next waveDrafting contracts, proposals, client reports and forms is moving fast, but a person still signs off.
Typical capabilities: Proposals, client reporting, RFPs, regulatory filings
Organisational knowledge
SlowestManual work resting on unwritten rules, relationships and judgement sees little real AI use yet. Only a third of investment firms report meaningful ROI, and data gaps slow everything down.
Typical capabilities: Exception handling, complex trade operations, custody breaks
Ground rules
- Work built mostly on speech or maths starts at High; Transformative when adoption is visible.
- Work built on reading or writing documents is Moderate to High and flagged 'next wave'.
- Work resting on organisational knowledge stays Emerging until real use is evidenced.
- Drop one level for heavy regulation or mandatory human sign-off; rise one when large firms visibly build it in-house.
- New entrants count only as evidence of investment, never as the score.
The score
Each capability gets four scores from 0 to 3. Today's real AI use counts double; limits subtract.
- Work done or changed by AI today (×2)How much of the actual task is now done or shifted by AI?
- Where it's heading (3–5 yrs) (×1)What does authority research expect next?
- Investment flowing in (×1)AI funding, launches and acquisitions aimed at this work.
- Limits (×-1)Regulation, human sign-off or data gaps that slow AI down.
Total 3 or less: Emerging · 4–6: Moderate · 7–9: High · 10+: Transformative.
Sources
- Kitces / InvestmentNews — AI notetakers fastest-growing AdvisorTech category (~18% adoption)
- T3/Inside Information 2025 — 41% of advisors use generative AI tools
- EY 2025 — biggest GenAI savings in compliance, risk and IT
- SIFMA AMG & Sia Partners 2025 — documents, alts reconciliation, corporate actions
- Oliver Wyman / AWS — GenAI first lands on documents and repetitive work
- Cognizant — written and spoken work, manual document review most exposed
- UK Finance / Accenture 2025 — complaints agents, KYC, software delivery
- Publicis Sapient / ThoughtLab 2025 — only a third of investment firms see real ROI
- TABB / InterSystems — capital-markets AI live but uneven, held back by data
- Goldman Sachs builds Claude agents for trade-accounting reconciliation and onboarding (2025–26)