Generative artificial intelligence or GenAI is steadily becoming part of everyday workflows in the finance industry, shaping how research is conducted, risks are assessed and decisions made. As the pace of adoption accelerates, it’s important for the industry to clarify how the technology can be used responsibly while preserving human judgement.
The emergence of domain-specific tools such as Claude for Financial Services reflects a broader shift in the industry. Rather than generic large language models, firms are increasingly looking to adopt GenAI systems that are tailored towards domain knowledge, specialised datasets, unique workflows, and industry-specific governance requirements.
For research and investment professionals, this raises practical questions, such as which parts of the research process can be automated, where oversight is critical, and how professional skill sets must evolve.
Investment professionals have always been under pressure to perform. In the past, this may have meant covering more companies, responding faster to market developments, and integrating different sources of risk into investment decisions.
All of these tasks take on new meaning as the profession rides on the wave of GenAI and digitalisation.
Tasks ranging from interpreting vast and unstructured datasets, identifying patterns and flagging anomalies, and optimising investment models with different correlation, can now be automated in a way that allows incorporation of different datasets and more frequent updates, making it less labour and time intensive. This is especially important during periods of heightened volatility and market dislocations.
A recurring theme in CFA Institute’s conversations with investment firms is uncertainty about AI’s impact on careers. Will it reduce demand for traditional investment roles? And if machines can perform much of the technical work, what will be the core value of the investment professional?
Use cases
Evidence from CFA Institute research shows that investment professionals typically adopt a multihoming approach, combining multiple data sources and analytical environments to offset the individual system limitations and strengthen insight generation.
These increasingly embed AI tools, which are used to support synthesis of information, summarisation, detection of anomalies, and workflow acceleration, reflecting our research findings that AI is becoming a routine part of investment data processing.
Such an approach is particularly helpful in a region like the Asia Pacific where market structures and disclosure standards vary widely.
Retrieval-augmented generation or RAG techniques automate labour-intensive tasks such as reviewing financials, scanning corporate governance filings, collecting sustainability data, and synthesising news flows across portfolio companies in various sectors and geographies. The results are clearly structured to support downstream analysis and insight generation.
Slow and resource‑intensive work can now be scaled efficiently, shifting staff time away from data extraction and formatting towards higher‑value analytical and decision‑support tasks.
AI-generated results are just one piece of the puzzle. Once AI completes its work, human expertise comes into play.
For investment professionals, their value lies in what happens next: reviewing unusual findings, validating assumptions, assessing materiality, and determining when issues require further attention. These are all tasks that rely heavily on human judgement and insight.
For example, in the corporate governance context, interpreting whether a company’s actions require escalation still requires human judgement and engagement with management.
And in the sustainability context, professionals must combine data literacy with mastery of sustainability and be able to navigate evolving definitions, regulatory expectations, and differences between domestic and international reporting standards.
While AI enhances efficiency, it does not remove the need for local knowledge, a healthy dose of scepticism, and contextual understanding.
Next phase
The next phase of AI adoption is likely to involve agentic AI, which are systems capable of planning, sequencing and executing multi-step tasks across different tools with limited supervision.
In investment management, this could include automated monitoring of portfolio companies for regulatory developments, controversies related to environment, social and governance, or earnings surprises, with professionals applying judgement to interpret and act on the insights generated.
By automating this continuous, multi‑stream monitoring, agentic AI can help ensure that relevant developments are captured promptly and consistently without the need to manually scan fragmented sources.
However, responsibility does not shift to the machine. Investment professionals remain accountable for interpreting outputs, understanding model limitations, and exercising judgement in how insights are applied.
As with any advanced technology, the effectiveness of agentic AI depends on human oversight, explainability and auditability, and a disciplined process for reviewing and contextualising the results it produces.
Judgement and ethics
As GenAI absorbs more mechanical elements of analysis, the comparative advantage of an investment professional increasingly lies in judgement, ethical reasoning, the ability to determine data quality, and the capacity to navigate trade-offs in uncertain situations. This includes understanding how data is generated, recognising bias, and making decisions that align with the long-term interests of clients.
Stewardship is a clear example. In the Asia Pacific region, effective stewardship requires more than tracking voting statistics; it involves understanding corporate culture, incentives, and engagement outcomes. GenAI can support analysis, but humans are responsible for engagement decisions and voting.
Similarly, integration of sustainability factors in the investment process involves navigating evolving standards and trade-offs. Managers must decide how to treat incomplete disclosures, how to assess credibility, and when to engage or escalate concerns. These decisions cannot be automated.
Geopolitics adds another layer of complexity. GenAI can aggregate information on sanctions, trade flows or political developments, but assessing second-order effects on supply chains, capital access, or regulatory risk requires experience and contextual understanding, particularly given the region’s diverse political landscape.
AI+HI
The future of investment analysis will not be defined by humans versus machines. It will be AI+HI, i.e. combining AI and human intelligence.
For investment professionals, this means embracing AI as a productivity and insight generation tool while strengthening the skills and mindset that technology cannot replicate, or the four “Cs” – curiosity, creativity, critical thinking, and continuous learning.
CFA Institute’s ongoing research aims to support investment professionals through this transition, helping them to understand how GenAI reshapes professional practice while reinforcing the importance of ethical reasoning, stewardship and accountability.
GenAI will undoubtedly change the investment management industry. But it does not change the core responsibility of the professional: to apply judgement, act ethically, and make decisions that serve clients over the long term. In an AI-enabled world, that responsibility is not diminished – it is amplified.
*Mary Leung is senior adviser, capital markets policy, Asia Pacific, at CFA Institute.
