A global financial institution can match 98% of its transactions and still face thousands of unmatched exceptions each day. For the staff investigating those breaks, the remaining workload determines how much value automation delivers. Compliance reviewers need to trace the data behind each exception and demonstrate the reasoning behind each resolution.
Founded in 1994, AutoRek provides reconciliation and financial control software to banks, payment providers, asset managers and insurers. Scottish Equity Partners invested in the business in 2021 to support product development and international expansion. Today, the company is processing billions of transactions for more than 100 global clients.RegTech Insight spoke with Tony Livesey, Chief Technology & Product Officer at AutoRek, to understand how the company is applying AI across its core platform and the recently acquired Grath business.
More than 30 years of experience in reconciliation and financial control have given AutoRek significant domain expertise, alongside reusable solutions and configuration patterns. Rather than replacing effective deterministic processing unnecessarily, the company is applying AI to augment existing capabilities and improve how teams build, operate, investigate and resolve reconciliations.
The decision about where to deploy AI comes down to economics and the value it delivers for a particular use case. Livesey notes that conventional software can often produce the same answer as a large language model at a fraction of the cost and deliver the necessary response times at scale. As he puts it, “AutoRek uses AI where we see value from it, not just for the sake of it”.
The diversity of the AutoRek customer base results in a complex set of data types including PDFs and spreadsheets containing embedded XML. Handwritten notes coming in from the insurance segment take unstructured data handling to another level. AutoRek combines AI, conventional algorithms, existing integrations and third-party services to parse and format those inputs.
ARIA is AutoRek’s embedded intelligent experience for reconciliation teams, with capabilities tailored to operators, managers and developers. Its matching assistant identifies similar patterns among outstanding exceptions and proposes matches for users to review, helping teams investigate and resolve breaks that remain after established rules have processed routine transactions. Users can also ask questions in natural language, retrieve relevant product guidance and get help interpreting errors. Livesey emphasises that users retain the decision: “It always suggests and the customer then decides whether they want to accept that or not.”For managers, ARIA provides access to reconciliation schedules, workflow status and errors, helping them identify processing bottlenecks. For developers, it analyses data feeds, suggests configuration logic and highlights departures from recommended practice when building reconciliations. The July 2026 update extended automated rule generation for complex matching scenarios and added oversight and explainability features. AutoRek has reported a 95% reduction in time spent evaluating manual matches, a 90–95% reduction in the time needed to create matching rules, and configuration of routine reconciliations in under 30 minutes. These are specific published results, rather than universal outcomes across all customers or use cases.
AutoRek’s enterprise architecture provides isolation, security and control for institutions with demanding requirements around data protection and compliance. Smaller firms with simpler reconciliation needs may opt for a shared, multi-tenant cloud service in return for lower costs and faster implementation.
The recent Grath acquisition broadens the ways customers can access and use AutoRek’s reconciliation capabilities. Together, AutoRek, Grath and Topa support a range of approaches, from enterprise software and software-as-a-service (SaaS) to capabilities that can be embedded directly into customers’ own applications and workflows. Grath’s multi-tenant platform serves firms seeking faster deployment without operating their own infrastructure. Its use of AI to generate matching rules can also reduce implementation work and help AutoRek serve customers with simpler requirements at a lower price point.
AutoRek presents governance, auditability and human oversight as requirements across its offerings. Livesey explains what that means within a reconciliation workflow: users must be able to trace an unmatched item to its cause, inspect adjustments and understand the evidence behind a proposed action. Finance and compliance teams need that history to challenge recommendations. As he puts it, “the drill back into why the model’s made that recommendation is important”.
Topa extends these capabilities to firms that want to incorporate reconciliation functionality into their own applications and workflows through application programming interfaces (APIs). This gives customer teams another way to access AutoRek’s capabilities while retaining control of their own systems and architecture.
Livesey sees an opportunity for customers to develop their own interfaces and workflows using AutoRek’s matching, financial control and data lineage capabilities. AutoRek is exploring how to make more of its platform available to those customers, reducing the infrastructure and controls they would otherwise need to build and maintain.
Subscribe to our newsletter
