Technologies based on A.I. are helping create the next generation of automated reconciliation platforms — potentially transforming Ops from a back-office function to strategic partner in the trading room. But are firms ready for this quantum leap?
Reconciliation and exception management processes are converging with automation and artificial intelligence (A.I.) technologies and the union could be much more than an upgrade from manual processes. If managed correctly, it could be a quantum leap forward resulting in Ops reborn as a forward-looking trend tracking system for the front office.
Yet adoption faces significant hurdles. AutoRek’s recent survey-based report, “The Tipping Point for Asset Management Operations: The Rising Costs of Operational Complacency,” which is based on securities industry surveys, finds that firms have “heavily manual environments.” FTF’s own polling via a recent webinar, “Reconciliation Reset: the Shift from Liability to Operational Excellence,” finds that 42.1 percent cite manual processing inefficiencies as their biggest recs challenge. At the same time, 71.9 percent say they use a combination of reconciliation methods, complicating matters.
The silver lining in the webinar polling was that 52.5 percent of respondents are exploring A.I. for reconciliation, and an impressive 11.9 percent are currently using A.I.-based technology solutions. Yet nearly half, or 42 percent, cite trust and transparency concerns as top barriers to A.I. implementation.
Yet there is excitement about what may come.
New Systems, New ROI Metrics
Before firms leap, they will need to clarify their key performance indicators (KPIs) and return on investment (ROI) benefits and allow themselves to see beyond the improved data quality to the hidden values of applying A.I. to recs will bring.
A parallel takeaway exists in the financial crime-fighting disciplines of ‘know your customer’ (KYC) and ‘anti-money laundering’ (AML), which are essential to uncovering illegal activity and regulatory noncompliance. A.I. and automation can ease KYC efforts to verify client identities, and AML investigations would greatly benefit from A.I.-based analytics that identify violations and suspicious patterns in policies and procedures intended to protect a firm.
Roger Leahy, sales director, head of asset management for the U.K. and Ireland at Fenergo, cites Fenergo’s launch last year of its Intelligent Document Processing (IDP) solution, an A.I.-based, software-as-a-service (SaaS) document management system. The goal was to allow customers to “drop one or more documents into Fenergo CLM and let the system take care of the rest,” which entails document classification, data extraction, document linking, and completion, according to Fenergo. IDP uses large language models (LLMs).
“We’ve seen, on aggregate, the clients who have used it have a reduction in time spent on manual document handling of 72 percent. That’s the best proof point we’ve seen in terms of leveraging intelligent document processes,” Leahy says. The LLMs “remove the manual handling, the splitting, the categorizing, and the scraping of that data.” Leahy cited this example during an FTF webinar, “Transforming KYC & AML — A Smarter Era for Asset Managers.”
The time savings alone could justify bringing A.I. to recs, especially as 39 percent of firms report via the AutoRek survey that they expect increases in transaction volumes over the next two years. That means more reconciliations than an automated system can process efficiently without A.I. — making A.I. a requirement.
Murray Campbell, global head of buy-side product at AutoRek
“When firms adopt A.I. into their reconciliation processes, they need to be clear from the outset what indicators they’re going to use to measure the success of those features or those solutions that have been added into the process,” says Murray Campbell, principal product manager at AutoRek, in a recent FTF News video, “Can A.I. Fix Reconciliation? Murray Campbell on What’s Changing and Why It Matters.”
Campbell outlines six key metrics for measuring AI reconciliation performances:
- Ways to quantify and record the speed of the decisions returned via A.I.;
- The processing power required for each action and answer;
- Staff feedback about the solutions;
- A method to assess user satisfaction;
- The rate of successful matching suggested via the A.I.-based rules; and
- The rate of successful automated categorization.
As firms become more familiar with A.I., they will want “to enhance that final five percent or 10 percent of the automated reconciliation process,” Campbell says. “It’s looking at exception management, looking at other ways that A.I. can be used. It maybe just sits outside of the capabilities of a standard automated reconciliation tool and that’s where we’re really looking to use A.I. effectively to just enhance that overall process.”
Yet firms have resisted updating their recs until legacy Ops are near a breaking point, which was the focus of a recent FTF Exchange podcast, “The Efficiency Dilemma: Why Asset Management Ops Must Evolve.” But that resistance is giving way to change as operations teams are under pressure to decrease the cost of transactions.
But a key barrier in bringing down transaction costs is the sheer amount of data to be managed, adds Jack Niven, vice president, North America, AutoRek, who also participated in the podcast.
Jack Niven, vice president — North America, AutoRek
“Most of our clients work with four or five custodians, maybe more,” Niven says. “They might work with a couple of prime brokers, a couple of fund administrators. So actually, just all of that data, and getting it into a system or into a position where you can do a reconciliation, is a challenge in itself, let alone actually the matching element of it.” Add in trade breaks and investigations, and recs is hugely time-consuming.
Ops teams are under the gun to find solutions especially when the front office needs to launch new financial products to develop new revenue streams. The challenge for firms is that they need to make Ops changes without racking up transactions costs that are greater than their peers.
But overhauling a legacy recs infrastructure will require more than a cost per transaction argument or even the push to grow assets under management, Niven says. Executives will ask if they have to increase the Ops headcount to deal with greater levels of recs, new data types, and new financial products. Niven says these factors will force the question: “Is it time to put in a solution now? Or actually, is it something that we just want to continue doing manually for a period of months before we do get to that breaking point?”
Better Data Quality: The First Benefit
Nobody wants a breaking point but everyone wants clean data. Bringing automation and A.I. to recs could ease multiple data management and maintenance efforts across the trading enterprise, according to AutoRek’s report. “There is great value for firms in their data, however this is lost where processes are manual,” according to report.
The survey found the major data challenges are:
- Integration and compatibility woes: 39 percent;
- High volumes: 34 percent;
- Data entry errors: 33 percent; and
- Reliance on spreadsheets: 57 percent.
These data challenges will worsen as transaction volumes continue to rise. Data management and processing inefficiencies create operational risk. “The reliance on outdated methods for critical functions like reconciliations and compliance reporting makes firms more vulnerable to mistakes, security breaches, and regulatory non-compliance,” according to the report. The way out of this conundrum is with a new set of tools such as data analytics, automation, and A.I. “Peer organizations that invest in these technologies will gain efficiencies, improve decision-making, and position themselves to capitalize on emerging opportunities,” notes the report.
One caveat is that data quality concerns will vary by the type of firm, says Niven. “We work with fund administrators, asset managers, hedge funds, private equity firms, large investment banks,” he adds. “The use case ever changes.”
Yet the stubborn reliance on legacy technology for recs, especially at larger firms, causes problems for data ingestion. In fast-moving environments, the business side may outpace operational and technology capabilities, often resulting in patchwork solutions, such as relying on Excel, to bridge the gaps.
Finding Those Hidden Values
Recs fortified with A.I. could yield some talent retention and redistribution of labor that are not immediately obvious benefits.
Kristin Ferguson, senior director, enterprise investigations at TIAA- Nuveen
In another financial crime prevention parallel, Kristin Ferguson, senior director, enterprise investigations at TIAA- Nuveen, witnessed how cutting-edge IT can help firms not only retain talented staff members but also attract new hires especially if the IT excites the staff and eliminates manual processes. (Ferguson mentioned this hidden benefit while participating in the KYC & AML webinar.)
Initially, Ferguson and her cybersecurity team experienced the loss of top talent as they took positions where they could develop A.I. use cases and implement them. “Those are the individuals that are really being sought after right now,” says Ferguson, adding that her team went “back to being very manual and being very fragmented.”
However, there was a turning point when TIAA-Nuveen saw the benefits of A.I. “You’ve got individuals that have boots on the ground They’re getting excited about the fact that now they get to be involved [with A.I.] and that’s attracting individuals. You have the push and the support across the organization. So, now you’ve got IT and cyber folks that are coming back in and, and we’re actively trying to attract those individuals.” The firm is reversing the talent loss caused by frustration with manual systems.
“The excitement over A.I. is kind of trickling down at all levels. Because even if you are not one of the individuals that’s proposing use cases and, and developing the A.I., you have the excitement of ‘Okay, how is this going benefit me?’ When you get excited, that’s another piece that makes it easier to bring in top talent. I mean, how many times have you heard or you’ve talked to somebody who’s raved about an organization? That word of mouth spreads,” Ferguson says.
Chandrakant Maheshwari, senior model validator at Flagstar Bank
Part of that excitement spurs the reshuffling of Ops and recs roles, says Chandrakant Maheshwari, senior model validator at Flagstar Bank, who took part in the FTF webinar on KYC & AML. Maheshwari says that LLMs, key to A.I.’s progress, will need strong governance to mitigate risk and for validation.
“We need to have the right guardrails associated with these models,” Maheshwari says. “Overall, the wrapper is governance, right? Policy around the right regulatory guidelines that need to be followed. We will see a huge change in structure of teams’ talents.”
AutoRek says that exploring technologies is a magnet for younger talent unhappy with manual processes. “Giving someone a new solution and asking them to implement it and then to teach other people how to use it, is a skill, right? And that’s how people can then level up in their careers. You’re not going to level up in your career doing manual processes. You’re going to get stuck,” he says.
Building the Business Case for AI Investment
Inherent in the decision to embrace A.I. for recs is the challenge of getting buy-in from the executives who will have to fund all A.I. transformations.
Ferguson, at TIAA-Nuveen, reports that her firm has a risk committee that reviews any proposed A.I. solution, developed internally or externally. The review process encompasses the feedback from legal, risk compliance, ethics, privacy, and fraud teams. “We’re all looking at it through our own respective lenses,” Ferguson says. “You can’t expect one person to be able to think through every element or every aspect.” By having so many different individuals on board who must put their stamp of approval on an A.I. solution is helping the firm’s policies move forward. In particular, it is helping the firm develop policies and best practices that will help sell A.I. to the executives.
Roger Leahy, sales director, head of asset management for the U.K. and Ireland at Fenergo
In addition to an inclusive review of the risks, firms will get buy-in from the right stakeholders if they also invite everyone to join the journey, Leahy says. The key is to get “involvement from not just your Ops or your business team but having your trust and security teams and your technologist in the room as well,” he adds. “Whenever you’re going through those transformation programs, whether it be more basic moving from on-prem to SaaS or whether it be implementing machine learning, A.I. technologies, it’s about being clear in your operating model and, and bringing everyone with you.”
Another tactic to consider is involving the regulators especially when it comes to building compliance programs for A.I.-based efforts, says Anne Marie McAvoy, who is CEO of Clovis Quantum Solutions and a founding member of McAvoy Legal Services. McAvoy participated in the FTF webinar on KYC & AML.
“I think that you may actually want to bring the regulators into that journey as well,” says McAvoy. “So as you’re developing it, bring them in and teach them how A.I. is operating in your company, what you’re doing, so that you get the buy-in from the regulators. Because without that, you’re not going to be able to function properly in the long run.”
The AutoRek report notes that while firms recognize the need for automation, many firms ultimately struggle with implementation, which suggest a communication gap between the IT and Ops teams and decision-makers.
Being as clear as possible about those operational benefits are key to getting the attention of the executives, Campbell says. “When firms adopt A.I. into their reconciliation processes, they need to be clear from the outset what indicators they’re going to use to measure the success of those features or those solutions that have been added into the process,” he says. He also urges firms to craft use cases for the design and deployment of A.I. for recs that everyone can understand and get behind. Stakeholders need to see that the goal should be to deliver success “against those particular use cases rather than looking for a big bang approach to fix your complete reconciliation process.”
Maheshwari at Flagstar Bank also urges firms to avoid a big bang. Instead, firms should begin their journey with low-risk, high-volume tasks that would favor structure over the need for lots of analysis. This is the basis for building trust and for making discoveries that can be scaled upward afterwards. The key is to pursue modest, explainable wins, rather than leaping straight into high-risk automation, he says.
From Data Processors to Market Intelligence
Small steps can lead to bigger ambitions such as transforming Ops teams into a new source of strategic intelligence for the front office — “the golden source of trend information,” says Jim Sadler, chief product, technology, and operations officer for AutoRek.
Jim Sadler, chief product, technology, and operations officer for AutoRek
“If you think about a finance operations team, if you could move that team from being kind of a back-office function that just has to get through a dataset in a certain time to remain in control or compliant, that team through the use of A.I. could quite easily become a frontline strategic asset,” Sadler says.
However, there are a few hurdles before that can happen: overcoming high match rate challenges, dealing intuitively with exception scenarios, and making recs and controls “a hygiene factor rather than an essential operational part of the business,” Sadler says. “That team could become the next source of strategic insight for the firm. And the reason for that is that the team sees every transaction. Everything that was done is in the data set that that team is looking at. So within that data set are industry patterns, market trends, customer behavioral shifts.”
A current Ops team configuration has “all of that data at its fingertips, not actually performing that well because they’re too concerned with controls and compliance,” Sadler says. “Whereas with particularly agentic A.I. or very intelligent A.I. systems, then the Fin Ops team could quite easily become the forward-looking radar tracking system that looks ahead to see what decisions could be made next and where the finance function and the business should focus next.”
As Sadler notes, this new strategic role for the Ops team will not happen overnight. Until it does, Ops staff should focus on the new values coming from A.I.-enabled recs.
“Certainly, anyone in finance operations leveraging intelligent systems should be immediately thinking about how they can redeploy the workforce to value creation. And again, you might think about these in terms of a KPI context,” Sadler says. “It would be a real shame if the only KPI that mattered was cost saving. It would be a real benefit if the only KPI that mattered was value creation, that you can now perform because you’ve unshackled your team from repetitive controls-based work.”
Need a Reprint?