Bio
Matchbook AI is a data intelligence SaaS platform that accelerates a company’s decision making by augmenting internal enterprise data systems such as vendor, procurement, customer and marketing info, to syndicated data from partners such as Moody’s, Dun & Bradstreet, Melissa and Zoominfo. It counts Google, P&G, Boeing and other mid-sized to Fortune 500 among its customers.
Situation
Matchbook's platform offers a data stewardship toolbox that requires customers' development teams to integrate using API keys in order to access its features. The platform includes automation capabilities for setting up confidence codes to determine which data matches to accept and which to review. Once validated, the combined data sets, known as mastered data, can be used for processing and generating insights.
To assist users with integrating and troubleshooting, Matchbook established a help centre with documentation with detailed video recordings and transcripts. However, accessing this knowledge may not always be instant and could involve time-consuming information discovery. In response to this challenge, Matchbook's CEO, Rushabh Mehta, tasked Alltius with developing an assistant that would be trained on these resources. The assistant would provide instant answers with clear how-to references to users and redirect them to raise a ticket in their HubSpot workbench if further assistance is needed.
Solution
Alltius and Matchbook's product and engineering team swiftly integrated an on-brand widget into the platform, leveraging Matchbook's extensive help resources to train the AI assistant. The assistant was accessible only to authenticated users within the Matchbook platform and was estimated to handle 90-95% of queries with accurate instant responses.
In cases where the assistant couldn't provide a satisfactory answer, a unique intent detection model was implemented to identify when users needed human support. Upon detection of this intent, users were given the option to raise a ticket. This would typically happen in the rare cases when the assistant didn’t know the response to the user query or the user was not completely satisfied with the self-help offered.
The assistant would then summarize the conversation, find user credentials, create a contact if necessary, or raise a ticket, all using secure API connections with Matchbook's HubSpot account.
Results
Matchbook successfully implemented an AI assistant to streamline their user support process, reducing the median wait time by 90% and enabling quick responses to up to 95% of user queries. This has freed up developers and agents by 20%, eliminated manual data entry in ticketing systems, and improved customer service efficiency by automatically summarizing conversations under the correct user records.
The setup process, usually handled by product managers, was completed effortlessly without the usual challenges associated with traditional rule-based chatbots.
Way forward
In the future, Matchbook intends to implement automatic transcription of recorded support videos and provide users with video excerpts using Alltius' platform. The pilot program has also initiated a partnership to develop highly personalized assistants for Matchbook’s own downstream clients to assist with data curation from numerous unstructured documents such as 10K reports and scanned images of paper documents.
