Conversational AI for self-care and customer support

Cas client

Corporate & investment banking

Conversational AI for self-care and customer support

À propos du client.

Corporate and institutional banking division of an international bank. The division serves 18,000 corporate clients and employs 36,000 staff across 55 countries.

What was your approach to the issue?

Talisker’s involvement in the mission was positioned at the interface between the business and IT teams of the Operational Excellence AI hub, responsible for the implementation of Amelia conversational AI across the bank’s various divisions. The work was organized around three main areas:

  • The first area focused on identifying use cases and building the associated conversations for the AI. This involved identifying the use cases to be implemented (workshops with the business, ticket analysis, etc.), writing functional specifications for the IT teams, and ensuring the link with the product teams responsible for user support on the client portal to guarantee the relevance of the responses provided (documentation, menus, etc.).

  • The second area concentrated on liaising with the IT teams and version tracking. The key actions included acting as the business point of contact for the AI development teams, conducting tests, and following up on feedback to ensure successful production releases. The advisory role was also important to strengthen the autonomy of the bank’s Operational Excellence hub associated with the Amelia AI solution.

  • The third area targeted language understanding and training the conversational AI. This required a strong understanding of business needs to build the dedicated intent architecture (categorization of request types) and the validation set (VDS), which was then used to train the AI.

What was the key to your success?

The strong involvement of the business Product Owner for conversational AI within the Corporate Investment Banking division, as well as their significant drive to engage sponsors, were undoubtedly key factors contributing to the success of the mission.

Our shared experience with Amelia also helped streamline communication and bring in the necessary expertise at the right moments to accelerate project development.

The in-depth use of the existing, comprehensive documentation for clients was also a determining factor in promoting self-care by quickly providing quality content to customers.

Where did you start from?

The project was initiated in the context of significant congestion in the customer service department. Employees are frequently tasked with handling recurring simple issues, and the responses provided do not aim to educate clients on how to use the platform.

The support teams are fragmented, and a user request can take up to 7 days to reach the appropriate service desk, resulting in long processing times and a degraded customer service experience.

Conversational AI is already being used in another division of the bank, which chose to invest in the Amelia solution. The Corporate Investment Banking division aims to build on the encouraging initial results achieved at the group level to develop a conversational AI solution for its clients on its online platform.

What were the results?

The intervention led to several key results, achieved step by step through the delivery of different versions of the conversational AI for the Corporate Investment Banking division’s client portal.

Version 1 (V1) enabled the AI to redirect tickets to the appropriate service desks based on the subject of the request and the user’s country. At this stage, the AI was available internally only (proof of concept).

Version 2 (V2) provided a first level of self-care to users, based on the available user documentation (videos, one-pagers, user guides, etc.). The AI was then opened to the first pilot clients, with 300 pilot users.

Version 3 (V3) allowed the AI to autonomously respond to certain questions and assist users in opening tickets by automatically retrieving their payment information. The number of pilot users was then extended to 600 users.

Version 4 (V4) marked a turning point in product maturity, as the AI was able to understand more question types (40 intents) and conversations became more natural. This version also introduced the first phase of a mass rollout for EMEA clients, with a target of over 200k conversations per year.

The organizational efforts led to the creation of a product team consisting of 3 business resources and 4 dedicated AI resources for the Corporate Investment Banking division.