Investment organisations have spent years building the technology and data foundations that support the investment process. The next challenge is how to use those foundations to create an ‘edge’.
That was the focus of a discussion at the recent 11th Investment Data and Technology Summit in Sydney, where the panel considered how AI could reshape investment operating models and where organisations should focus their efforts next.
The expectations of what AI can achieve are significant and go far beyond simply reducing manual work and lowering costs and operating expenses. An audience poll at the summit found that delegates expected far greater economic value from capacity increases and improvements in outcomes as a result of using AI than that they expected to see cost savings – though the cost of token usage might have had something to account for that too; a subsequent poll revealed that just eight per cent of the audience had measured a reduction in cost as a result of using AI.
Asked where they expected AI to deliver the greatest economic value in their operating model, 59 per cent selected greater capacity – doing more without proportionally increasing headcount. A further 33 per cent selected better outcomes through improved decisions, speed and responsiveness, while just seven per cent selected lower costs through reduced manual work and operating expense.
From system of record to system of edge
The conference heard that much of the investment technology infrastructure built over recent years remains focused on what one panellist referred to as the “system of record” – the foundational book of record systems that support the investment process after investment decisions have been made. This include structured data held in systems such as ABOR and IBOR, but it doesn’t necessarily capture other potentially valuable information that is contained within reports, research and other unstructured sources.
The panellist said that, looking to the future, the priority is now to build a “system of edge”: those capabilities that could help to improve investment decision-making and ultimately differentiate funds from their peers.
To support those objectives, organisations need access to those additional valuable data sources that are not captured in the systems of record and this is where current technology is making significant inroads.
Examples discussed included agents carrying out research and investment analysis before returning their work to a human to assess whether the output makes sense, though most applications at this stage are still focused on individual productivity rather than enterprise-wide transformation.
The interface is changing
Looking further ahead, one view put forward was that people could increasingly access existing investment systems through AI interfaces rather than conventional applications.
In the medium term, this could mean users increasingly interacting with systems of record through tools such as Claude or Copilot. Over a longer period, the way workflows currently move information between applications and analytical processes could change substantially, potentially leaving a data foundation with an AI-based interrogation layer over the top.
One proposed architecture for the emerging system of edge consisted of a data layer, a context layer describing and governing that data, an intelligence layer containing models and analytical tools, and a user experience layer through which people interact with the system.
From routine to judgement
While that wouldn’t necessarily change the systems that are currently in use, it would certainly impact on the way that people interact with those systems.
Panellists also flagged routine activities such as reconciliations, spreadsheet production and presentation-building as increasingly candidates for automation. The human role could instead move towards interpreting analysis and applying judgement, they said, though the distinction would depend on the task.
Where precision and consistency are critical, human oversight would be essential. However, in other situations, such as generating ideas or challenging existing thinking, AI could play a useful role.
The panel also cautioned against assuming that every existing process should become AI-enabled.
A monthly performance reporting process was used as an example. If a process is already highly automated, predictable and inexpensive, introducing an AI system may simply add complexity and potentially cost without delivering additional value.
The starting point should therefore be the business problem and the value that a new approach creates, rather than the availability of the technology.
That applies equally to decisions about technology and sourcing. The panel noted that organisations have historically made outsourcing and technology decisions that could leave them tied to particular providers and architectures for many years. Increasingly, there is an opportunity to build more flexible environments that can evolve as technology changes.
One organisation described taking a deliberately clean-slate approach to its cloud environment, removing legacy technology rather than continuing to patch and extend it. The resulting environment provides users with a controlled platform through which they can access new applications and AI capabilities, although keeping staff skills aligned with the pace of change remained a challenge.
The emerging operating model is therefore not simply about replacing existing systems or reducing headcount. It is about deciding where AI can genuinely improve the investment process, how proprietary data can create an advantage, and how people, technology and processes need to evolve around it.
