Photo of Sharon Fang and Dennis Crispin, Microsoft.
Microsoft’s treasury team has built a robot, named Leonard, that has enabled it to optimise resources by dealing with stakeholder questions. By using AI, Leonard is constantly learning and will increasingly add value to Microsoft making it an indispensable member of the treasury team.
Senior Manager Ecosystems
Founded in 1975, Microsoft is the worldwide leader in software, services and solutions that help people and businesses realise their full potential.
Building the Digital SME
Microsoft’s Worldwide Payment Solutions (WPS) team had a simple yet time consuming issue: it was receiving too many queries from the business over email that required a manual response from the team.
Therefore, in an effort to optimise resources and make it simple for its stakeholders to learn more about what the team does and how it does it, Microsoft set about finding a way to automate this process.
The team decided to employ a chatbot as part of the group’s digital transformation strategy.
The strategy was simple, build a bot that could start with a simple question and answer capacity and go from there. Using the Azure Bot Service, the team’s Finance Rotational Analyst was quickly able to transform an antiquated and clunky Excel-based FAQ library into a fully functional Q&A chatbot.
The initial version of the bot, called Financing Assistant and nicknamed Leonard, used natural language processing (NLP) to interpret questions from users and respond with useful answers about Microsoft’s financing programs, internal tools, and policies. Active learning – an artificial intelligence component that helps ensure proper responses are given and keeps the bot continuously learning – was built into the bot as well.
In just a few weeks the WPS team deployed Leonard as an internal contact in Microsoft Teams, the chat-based workspace that combines workplace chat, meetings, notes and attachments.
The team then integrated the bot into its new financing approval workflow and documentation tool single point of contact (SPOC) to give users an additional channel in which to interact with the bot. As part of the SPOC integration the bot was trained with additional skills such as viewing personalised reports, checking the status of financing requests, as well as cancelling requests. Additionally, the chatbot learned to communicate with other bots, specifically the bot of its sister team Worldwide Credit Services and users also gained the ability to communicate with the bot through speech by using Azure Cognitive Services. As some of the data in the bot may be sensitive, Azure Active Directory was utilised so the bot knows who it is chatting with and can direct responses accordingly.
The bot can also help users de-stress by telling them workplace-appropriate jokes. Interestingly, the joke feature gets users interacting with the bot and then they quickly become comfortable with more technical and policy scenarios.