Anmol Verma highlights the evolving role of AI agents in finance, advocating for systems that interpret, respond and learn, transforming how investors manage their portfolios.
Investment platforms have become far more capable over the past decade, but they still rely on the user to decide what matters, when to act and how to interpret the flood of market data. Anmol Verma, founder of the AI wealth management platform Finn, argues that the next step is not simply better software, but systems that behave more like colleagues: watching, interpreting and nudging investors towards the next sensible move.
In a recent essay, Verma said the point of agentic finance is not to make investors less involved, but to give them more intelligence at each step of the process. “The promise of agentic finance is not that investors make more decisions,” he says. “It is that they can bring more intelligence to every decision, without being constrained by how much information a human can individually track and process.”
That shift matters because most financial automation today is still narrow in scope. Portfolios can be rebalanced, risk can be monitored and trades can be triggered by preset rules, but these systems usually respond only to instructions or specific events. By contrast, AI agents are being designed to interpret context, weigh changing conditions against an objective and help coordinate a response rather than merely execute a task.
IBM has described AI agents in finance as tools that can manage workflows, coordinate functions and provide real-time insight with less direct oversight, freeing professionals to focus on strategic judgement. Oracle, meanwhile, recently unveiled 22 Fusion Agentic Applications in London, saying they are intended to push AI deeper into enterprise systems across finance, HR, supply chains and customer service. Anthropic has also launched specialised agents aimed at reducing some of the most time-consuming work in financial services, from banking to insurance.
The challenge, however, is not only technical but also practical. A system that reacts to every market move would create noise rather than value, so the real test is whether it can distinguish signal from distraction and understand when no action is needed. That requires a sense of context: the same company update may be irrelevant for one portfolio and material for another, depending on the investor’s goal, time horizon and thesis.
Verma expects adoption to unfold gradually. Agents may first summarise what has changed, then suggest possible actions and later take on more of the operational work around a decision. The long-term prize is a learning loop in which each recommendation, override and outcome helps the system become more useful over time. As that happens, investing could move from a model built around reactive tools to one in which software acts continuously in the background, helping investors focus on judgement rather than data handling.
Disclaimer: This article is intended to inform and educate, not to recommend or endorse any financial product, investment or strategy. Please consider your own financial circumstances and seek professional advice where appropriate before making financial decisions.
