Reconciliation Suite (RS)

An end-to-end machine learning-powered reconciliation software application

 

Reconciling Data Is Costly and Cumbersome

Reconciliation processes in FIs have become more challenging due to a combination of factors like increasing process complexity, high transaction volumes, a large number of data sources, ever-changing regulatory requirements and the emergence of new asset types and structured deals. Regulations such as Basel III and MiFID II mandate increased depth and breadth in the reconciliation process. Manual processing and rules-based solutions are poised to become costly and prone to numerous errors in the long run. The problems with current reconciliation processes across the globe (rules-based or robotic process automation (RPA) solutions) are:

  • Need for granular predefined rules or exact steps to reconcile transactions
  • Inability to reconcile complex transactions, resulting in exceptions or breaks
  • Exceptions form 10-15% of total transaction data but take 70-80% of an analyst’s time in investigation and resolution.
  • Exceptions handling and complex matching require human judgement beyond a basic set of rules, making the process costly and time & resource-intensive.
  • Some exceptions remain unresolved due to their complex nature or insufficient information causing aged breaks. This increases the risk of regulatory non-compliance.

Most solutions in the market support simple matching but they provide minimal support on complex matching and exceptions handling.

Tookitaki RS: Better Match Rates, Accurate Break Resolution

  • Generates ~90% match rates for general matching cases with minimum human intervention through our proprietary supervised pattern detection approach.
  • High coverage on exception handling with 95%+ accuracy and up to 70% reduction in exception investigation and resolution time.
  • A thorough audit trail that can explain the cause of a break in detail, helping FIs with actionable steps and explanation across model outcomes.
  • Sustained engine efficacy through advanced self-learning algorithms that incorporate incremental changes in data.
  • Easy integration to the FI’s existing and future up/downstream systems through built-in connectors and REST API interfaces.

RS Modules: Providing Robust Coverage and Scale

  • Matching Module: It auto-generates rules for general matching cases using supervised machine learning techniques.  It also handles complex match cases, helping overcome current matching inefficiencies through our graph-based complex pattern matching engine.
  • Exceptions Module: It uses semi-supervised learning to accurately detect breaks and recommend adjustment amount for break resolution. The approach allows detecting new unknown exception cases which would be difficult to manually investigate. Our proprietary semi-supervised techniques combine historical understanding and current behavior to recommend better handling of exceptions.