Tools, Trust, and Talent

Executive Summary

In AIMA’s previous report, Charting the Course: Lessons from AI Leaders, we examined how the firms leading the first phase of Gen AI adoption were preparing for the technology’s arrival. The central lesson was that progress depended not only on access to tools, but on leadership, governance and the organisational confidence to learn safely.

This year’s report considers how those lessons are being applied in practice. Gen AI adoption is now widespread across the fund management industry, but its integration remains selective. What earlier AIMA studies identified as front-office potential is now beginning to appear in practice: firms are using Gen AI widely in the research and analysis supporting investment decisions, but far less in portfolio construction. Caution also remains where sensitive data and regulated responsibilities are involved.

The report examines this changed landscape through three connected themes: Tools, Trust and Talent.

The tools available to firms are becoming more capable and the environments in which they are used more controlled.
Trust continues to define the boundaries of adoption, particularly where accuracy, explainability and accountability matter most.
The early talent effects are appearing less through widespread job losses than through changes to tasks, skills and team composition.

Scale influences each of these developments, but does not determine them. Larger managers have more scope to invest in private infrastructure and integration, yet the research does not show that access to greater resources automatically produces a broader spread of value. Many of the foundations for progress—clear ownership, approved tools, defined data boundaries, staff education and human oversight—are organisational choices before they are major capital investments.

Taken together, the findings point towards a more disciplined phase of adoption. The priority is to establish sufficient ownership, control and human oversight to test appropriate applications safely, then use evidence to determine where Gen AI should stop, adapt or scale. This gives firms of every size a practical route forward: readiness need not be perfect, but it must be sufficient for safe learning.

The Key Findings that follow show how this transition is reshaping the tools firms use, the trust they require and the talent they need.

Key Findings

A multi-model market is emerging behind controlled access. Claude is now the most widely permitted public LLM, at 67%, and the most commonly used model across six of seven task categories. Investment assistance remains fragmented, with 44% reporting no clear leader.
Firms police data boundaries more often than they prohibit functions. Sensitive-data restrictions are the leading control, selected by 69% of respondents. 31% identify enterprise or private-model-only use as a principal restriction, while just 3% report no restrictions.
Reported value is widespread and strongest in investment research. 89% identify at least one area in which explored Gen AI use cases have delivered the expected value. Investment research leads at 74%, followed by operations at 50%, technology at 48% and compliance at 46%.
Gen AI is helping firms build the investment case more often than the portfolio. Only 31% use it at least occasionally in portfolio construction, including 7% who consider it a core part of the workflow. This compares with 77% for investment-memo preparation, 73% for company or asset analysis and 71% for idea generation.
Trust sets the pace of wider adoption. 60% consider explainability important or essential for AI-assisted investment decisions. Accuracy and reliability, at 58%, and data leakage, at 51%, are the leading constraints on further use, well ahead of cost at 8% or internal culture at 7%.
Scale changes the operating model more than the breadth of value reported. Private or mixed-model adoption rises from 45% among sub-$1bn managers to 78% among +$10 billion managers. Yet among firms reporting value, the breadth of value reported is similar across AUM bands.
Respondents expect disruption to reach analytical groundwork before final judgement. Administrative roles and analysts are selected most often among the roles considered most at risk, at 43% and 41% respectively. No respondent selects portfolio managers.
Team composition is expected to change before total employment. Only 7% expect headcount to fall over the next 12 months, while 14% expect it to rise. 19% anticipate greater technical hiring, compared with 16% expecting reduced junior hiring and just 1% expecting reduced senior hiring. The effects may also extend beyond firms, with 38% reducing or expecting to reduce their reliance on service providers, although most of that change remains prospective.