SOAR (System Orchestration, Automation and Response)

Inspiration

Reconciliations done manually is time consuming and tedious. Ideally, it should be fast and efficient, yet expediting this process still remains a problem. Once a person’s attention lapses, it will increase the likelihood of human error and the costliness of the errors. Reconciliation errors can cause the company major consequences such as suspicion for fraud and accounting lapses. High transaction volumes, multiple bank accounts, different transaction types, multiple currencies and various bank file formats worsens the problem.

According to a financial accounting survey by EY, up to nearly 60% of a financial department’s resources are spent on managing transaction-intensive processes. However, 95% of this effort is spent on transactions that already match, rather than problem entries that actually require attention, reducing the efficiency of the organisation and dramatically increasing the time it takes to close. This is the gap SOAR aims to close and improve.

What it does

We came up with the SOAR (System Orchestration, Automation and Response) to automate and ease the process of reconciliation. That will considerably speed up the process of reconciliation and provide users with more flexibility with their time working.

Machine learning is utilised in the main process of SOAR:

• Onboarding of data
• Reconciliation
• Automatic resolutions and alerts

The onboarding process takes a long time as different companies have different forms of storing. We can create value with SOAR as it takes out slow human-based processes in the initial data onboarding stage and replaces them with a machine that analyzes and teaches itself how to handle the new data. Furthermore, it will automatically check data quality within specified tolerances, generate exception reports, and output a file to be ingested by the firm’s accounting system.

For the process of reconciliation, an AI-powered matching engine will be used to maximize match performance and accuracy without the need for manual rules or updates. It will be able to process millions of historical transactions, display exceptions and mismatches, and suggest matching rules. With the help of anomaly detection and clustering machine learning, it will attempt to resolve as many mismatches as possible while flagging and alerting any that require other methods of resolution.

Using technology such as Artificial intelligence trained in anomaly detection, clustering and general automation, it is easier and faster to identify irregularities in reconciliation. It enables accounting teams to focus on key areas, thereby improving their effectiveness and efficiency.

User Experience

Ideally, SOAR should be given a testbed within an accounting department to test its viability and cost of implementation. From there, metrics and data can be collected on how effective SOAR is. With the database that SOAR creates, SOAR will become a self-sustaining solution where AI and Machine Learning led solutions can be a beacon to illuminate the foggy work processes of reconciliation. Although the initial investment for SOAR’s network might be high, but the potential returns scales even higher.

Challenges,Accomplishments and Learning points

We did not have any background in accounting at all and this was our first foray into the world of auditing, accountancy and finances. Studying and scrounging for pain points was extremely difficult as we could not relate to the problems provided,thus we had to study in depth and properly understand the procedures and terms used. Furthermore as 1 of us had internship and the other had national service, we had very little free time to even create anything. We managed to make a working technical proof of concept with no prior experience in data analytics which was a huge achievement for us. If we had better planning, more teammates with a much more diverse skill set would have come in handy. Time management would have allowed us to better polish the prototype and improve the write up and presentation as well as perhaps a more refined idea.

What is next for our product

Our plans include developing SOAR into a viable system to invest in as it potentially shortens, simplifies and speeds up the reconciliation process. Not only is it self-sustaining, it is also easy to implement and has the potential to be far reaching. Ideally as an industry solution it should have synergies with existing systems which we plan to integrate with external API sources as well provide its own Restful API’s to allow other software to integrate with it.

Built With

• ai
• api
• bootstrap
• css3
• html5
• machine-learning
• matlib
• python
• restful