Hyper Intelligent Automation Implementation by Brillio led to significant overall cost savings and cost optimization with 500+ FTEs saved
Intelligent Automation Project of the Year
The client is an American-based Multi-National Corporation providing leading financial technology services, globally. They were looking to drive automation throughout their organisation to ensure that tasks such as document processing, loan processing, cheque data, dispute/fraud documentation and audit documentation which would aid in the reduction of FTEs and ultimately, a reduction in their overall operating cost. Brillio partnered with the client in their transformation journey to digitalize their front and back-office tasks by co-strategizing workshops to assess and discover automation opportunities. Brillio proposed a roadmap to automate 170+ use cases helping client in reducing their headcount to the tune of 500+ resources. However, we have initially implemented 4 uses cases and we plan to implement a further of 50 use cases by Q4 of 2022. Similarly, rest of the use cases will be implemented every quarter, in phases. The detailed description of one such use cases is as follows:
High cost of resources
• Mundane manual effort
• Repetitive tasks
• Time consuming processes
• Lack of automation
Example- Often a merchant would return a cheque back to the customer which needs to be scanned and uploaded to the remittance manager application by the Customer Data Center team, before the customer receives it. When returning cheques back to customers, merchants must attach a digitally scanned letter/document stating the reason for such a return. However, these must be validated, and corresponding case would be updated in CRM-PeopleSoft. This cumbersome process needs an innovative solution to reduce manual effort and prolonged time for such validation and updating.
Solution for use case:
Brillio’s solution ensured that this problem was immediately and effectively solved through the automation of cheque data extraction and processing by combining RPA Technology (for workflow automation), AI/ML Model and Deep Learning models for:
• Data pre-processing
• Entity extraction (to extract meta data from doc)
• Build a reason rejection model (to build a semantic model to identify reason for cheque rejection).
Through this, the data extraction model from handwritten/digital document was made easy and further processing of data on BU applications was performed via RPA Bots, thereby ensuring reduction in time and mundane manual effort by cutting down on repetitive tasks.
Similarly, all 170+ use cases were identified, and solutions were developed to automate the same and implement them.
POC and Recommendation- Before implementation of use cases, Brillio provided a POC with open source and commercial tools to showcase capability around document extraction/OCR (to maintain accuracy for extracted document) and NLP along with an assessment on scalability of the client with regards to document processing, loan processing, cheque data, dispute/fraud documentation and audit documentation and recommended a change from RPA- BluePrism to UI Path (e.g., pdf splitter functionality ais better in UI Path and allows easy integration with the AI/ML model) to accommodate all the use cases identified.
Implementation phases of the use cases:
• Initial Process Assessment (IPA)- UI path for AI/ML, IBM Watson, Azure cognitive services along with open source and NLU.
• Detailed Process Analysis (DPA).
• Automation and User Acceptance Testing (UAT).
• Deployment - making this model available as a service for RPA to consume for implementing an end-to-end automation without any manual intervention.
• Hypercare, Sign-off and Transition.
• Warranty Support, FTE Release and Final Sign-off.
Risks faced: Since every project is prevalent in a dynamic environment, it is bound to have certain risks. Brillio ensured that these were identified at an early stage and established mitigation strategies to ensure smooth automation transformation for the client.
• Financial risks - cost overruns and requirement of further tools, licenses or change in environment scale and size.
• Operational risks – For example, real-time dynamic input data (in case there is a deviation from the trained data while developing the AI/ML model).
• Technical risks - Adaptability to changing system entities during RPA workflows development and deployment which could also result in sensitive data being processed by BOTs.
• Execution risks - Delays in establishment of connectivity/permissions for accessing environments from our network.
Benefits and impact:
• 20-25% reduction in Implementation Costs
• 0 or no manual errors and reduction in time delays due to it
• Overall cost savings and cost optimization with 500+ FTEs saved