Artificial intelligence (AI) has penetrated many areas, and the asset management industry hasn’t been spared. Seven in 10 asset managers now use artificial intelligence in their front office, up from about one in 10 a year earlier, according to SimCorp, a provider of investment management software and services for the buy-side.
Dean McIntyre, chief commercial officer at the firm, said AI adoption is “improving productivity, helping users to focus more on value-adding tasks”, being deployed across investment teams, research workflows and client operations.
Across the industry, there is an increasing number of AI use cases, and they are becoming ever more specific.
How the industry is using AI
At BNY, an internal AI platform is used across the business, including sales and client onboarding, said Gerald Rehn, head of EMEA distribution at the firm.
“Client onboarding steps have moved from days to hours,” he said. More than a quarter of its staff use the system daily, with many building their own tools on top of it.
But it goes further than that. At Man Group, AI is embedded in the research process, as Tushara Fernando, head of data and AI at the firm, explained.
“We can onboard and evaluate datasets faster than ever,” he said. “We’ve found that leveraging both humans and machines has led to diversifying and uncorrelated content.”
On top of that, generative AI is used to analyse company filings, financial reports and news flow, allowing investment teams to cover more securities and spend more time on portfolio construction and risk management.
Elsewhere, BlackRock’s multi-asset range My Map combines systematic (80%) and discretionary (20%) investing, with machine learning introduced selectively rather than reshaping the process.
Chris Ellis-Thomas, co-manager of the range, said: “We carefully introduce elements that lean into machine learning or large language-model (LLM) platforms. We won’t switch things on without a thorough understanding of how they operate”.
One of the more successful use cases for him is sentiment analysis.
“[AI] is a good way of digesting news sources and understanding broader economic and market sentiment,” he said, describing how these insights can feed into broader risk-on and risk-off positioning.
Other firms are pushing further into generating investable signals. Aberdeen has built a model that reads central bank speeches and quantifies how hawkish or dovish policymakers are, generating an AI-built index that “has a lead on government bond yields” and “could be used as a trading input”.
AI allows investment teams to screen more securities, test more ideas and expand the range of inputs that feed into portfolios. In wealth management, it is already being used to identify portfolios that may need rebalancing, allowing managers to focus on the accounts most likely to require attention.
Agentic AI is also starting to be used.
“AI-powered agents can take multi-step actions, monitor changing market conditions and trigger workflows without needing constant human prompts,” said McIntyre.
For investment teams, this means routine tasks can be automated rather than handled step by step, freeing up time for portfolio decisions and client work.
AI agents are emerging both at Aberdeen and BNY, where they answer portfolio managers’ questions, carry out analytical tasks, edit research content and automate workflows.
The limitations
But simply adding AI tools does not create an advantage.
“Genuine AI use means rethinking business processes, not just asking AI to summarise a document,” Fernando said. The key question is whether AI changes how teams operate or simply increases the pace of existing workflows.
The quality of the underlying data is another constraint. “AI delivers the most value when it boosts decision-making and efficiency. That requires a unified, governed data layer,” McIntyre said. Without that foundation, outputs risk becoming generic, particularly as many models are trained on similar pools of public information.
This is why firms are investing heavily in proprietary systems and internal datasets. Man Group, for example, has built what Fernando described as an “AI-ready” data environment so its tools can operate in the context of its own positions, research and investment processes. Without this, the outputs of large language models are unlikely to differ meaningfully across firms.
There are also limits to how far AI is allowed to influence decisions.
“We do not have AI as a direct decision-maker in our critical processes,” Fernando said. Models used in signal generation are subject to testing, review and ongoing monitoring before being deployed.
BlackRock’s approach reflects a similar caution. The result is a hybrid process in which AI supports decision-making but does not determine it.
Is AI any good at investing?
AI does little to change the fundamentals of investing, according to Nick Clay, manager of the Redwheel Global Equity Income fund. “Ultimately, it makes life more efficient, but it doesn’t replace making the final investment decision”.
While useful, it may also introduce new risks, for example, widespread adoption could lead to more uniform behaviour.
“AI didn’t make picking stocks easier. Valuation drives total return. If everybody uses the same tools, they end up in the same stocks and create bubbles,” he said.
“When everyone was taught the same way to analyse and value companies, they all bought the same ‘cheap’ stocks and those stocks stopped being cheap. If we all use the same AI to pick stocks, we will all end up in the same place and prices will be too high.”
Ellis-Thomas noted that investors are simultaneously focused on the scale of investment going into AI and the risk of disruption it poses to companies, creating volatility that can present opportunities.
Asset managers themselves know a thing or two about this, with a sector-wide sell-off in February, triggered by startup Altruist's decision to introduce AI in its tax-planning offering, fuelling fears that human advice could soon be automated at scale (something that Tim Levene, chief executive of Augmentum Fintech, said is an “inevitability”).
Independent of whether it’s AI agents or humans doing it, the underlying mechanism of investing is unchanged, said Clay.
“AI doesn’t change the fundamental dynamic of markets, which is overvaluation and undervaluation driven by human emotions,” he said.
“It may make data processing more efficient, but it doesn’t improve your ability to choose the right stock. That still comes down to understanding valuation and doing something different from the market.”