Here is a detailed summary of the video transcription in markdown format, broken down into sections for better readability:
Introduction
- The presenters, Chris, Jill, and Steven, discuss how NatWest, a leading UK bank, has personalized customer experience using Amazon Connect and AI/ML.
- NatWest is a FTSE 100 UK bank supporting 19 million customers across retail, commercial, and affluent business segments.
- NatWest has 9 contact centers handling 25 million calls annually, supported by 9,000 agents.
NatWest's Journey to Amazon Connect
- NatWest started its journey to Amazon Connect in 2019, initially piloting in their credit card center.
- After a lengthy business case approval process, NatWest formally stood up the program in January 2023 and migrated 66% of their agents by the end of that year.
- By October 2024, all 9,700 agents were on Amazon Connect, completing the migration in 1 year and 10 months across 22 migration events.
- Key benefits of Amazon Connect include access to rich data, speed of delivery, and the "pay-as-you-go" model enabling optimization.
Architecture and Migration Approach
- NatWest's architecture includes both PSTN and VoIP (from their mobile app) ingress routes into Amazon Connect.
- They use natural language understanding (NLU) to steer customers to the right journey, with over 1,700 intents defined.
- Identity and verification are handled through voice biometrics and call line identification.
- Data is enriched and stored in DynamoDB, with call details surfaced to agents through the Contact Control Panel (CCP).
- To enable a safe and reliable migration, NatWest leveraged the Telco provider's number translation capabilities to cut over on a number-by-number basis.
- They prioritized migration over optimization, first gathering data on the new platform before making improvements.
Data-Driven Insights and Optimization
- NatWest now has access to 25 million data points annually, including call transcripts and agent data.
- They use this data for sentiment analysis, identifying and reducing hang-ups, and supporting vulnerable customers.
- Sentiment analysis allowed them to drive up customer and agent sentiment through targeted data-led coaching.
- Analyzing hang-up data revealed many customers were getting stuck in the IVR, leading to improvements in their natural language understanding.
- Identifying vulnerable customers enabled proactive support and addressing agent abuse.
Leveraging AI with Amazon Lex and Connect
- NatWest experimented with Amazon Lex to provide real-time agent guidance, but initial results were mixed due to the complexity of their knowledge base.
- They refined their approach, reducing the number of articles and structuring the content to better align with Lex's language model.
- Upcoming AI initiatives include call summarization to save agent time and agent evaluation to ensure compliance and quality at scale.