KHOJ, an AI based search engine for all things HDFC Bank! Qorus Banking Innovation Awards 2022

Submitted by

HDFC Bank

Premium
30/09/2022 Banking Innovation
KHOJ is an on-cloud AI enabled search engine which answers queries of employees across all customer facing units providing complete & up-to-date information regarding the bank’s Products, Policies, Processes & People at the click of a button.
Innovation details
Country
India
Category
Analytics & Artificial Intelligence
Keyword
Operational excellence & efficiency

Innovation presentation

The bank offers a wide range of products and services (300+) that are governed by complex regulatory processes which are subject to regular changes as mandated by Reserve Bank of India (RBI). The frontline staff need to have all the information w.r.t. 4 Ps - Products, Policies, Processes and People (managing the products/processes) at their fingertips which is currently not available. Employees have to rely on numerous disparate sources (mails, presentations, portals created by different teams, etc.) to get the required information which may or may not be up-to-date. This leads to a lot of to and fro before any customer request/query is addressed correctly resulting in a delay in fulfilling customer requests and thereby increase in complaints. To overcome this problem, we created an internal AI based search engine, KHOJ that answers all queries and provides the most relevant information w.r.t. the 4 Ps. KHOJ is our own internal Bank wide AI based unified search engine for all HDFC Bank employees. It works on Natural Language Processing, therefore one can seamlessly search for information using conversational style communication as well. Example: In case a customer walks into a branch to open a Senior Citizen Savings Account. The Relationship Manager (RM) would either have to reach out to the supervisor or another colleague who would be attending other customers during that time or would have to search for all the information at the various sources as given in Exhibit 1. – Before KHOJ (refer document uploaded). With KHOJ, as seen in Exhibit 2. – After KHOJ (refer document uploaded), the RM can simply type the query in KHOJ search box and get access to all the latest information at one place saving time, efforts and thus efficiently meeting the need of the customer. We have leveraged the following new technologies to build the search engine: 1. Microsoft Azure Cognitive Service capabilities - Language Understanding Intelligent Service (LUIS), a cloud-based conversational AI service for natural language processing. 2. MS Office 365 Sharepoint as the content repository and as the search screen user interface. 3. PowerBI, an advanced business intelligence tool to present and track search usage. Refer Exhibit 3: Search Engine Flow Refer Exhibit 4: Natural Language Processing Architecture MS Office 365 Sharepoint: Microsoft Office SharePoint is a cloud-based service that provides a secure and productive file collaboration environment. We have used Sharepoint extensively to customise the following: 1. Content: Setting up a content upload process which differs from uploading People related masters and Product, Policy or Process related content. 2. Data Governance: Updating latest content or modifying content thus maintaining a version history of the files along with publish date, expiry date and audit logs for data governance. 3. Feedback: Collecting, storing and actioning upon user feedback. 4. User Interface: Designing the layout and defining the page structure for each of the 4Ps and presenting the results in a uniform manner across. LUIS: LUIS is a cloud-based conversational AI service that applies custom machine-learning intelligence to a user's conversational, natural language text to predict overall meaning, and extract relevant, detailed information. By defining entities, utterances and patterns, the model is trained to identify intent from the query entered. It then matches the intent with the content available and generates LUIS Score of Confidence. Basis the top score response, the search hits Sharepoint to generate the search results. The ranking of the search results is determined basis a combination of the best match and number of hits on a particular search result. This goes as an input into the model along with feedback to improve the model performance and subsequently the search results. Power BI: We have utilised Power BI as an analytics platform to unify data from many sources within the Sharepoint and LUIS environment and to create an interactive, immersive dashboard that can provide actionable insights and drive the usage of KHOJ.

Want to keep reading?

Become a Qorus member to get access to all our innovations

Interested in learning more?

Qorus has a library of almost 8,000 innovation case studies across critical areas like customer experience, sustainability, marketing & distribution and more that can be used to inform your decision-making.

Related Content