Customer service calls, turned into insight
A speech-to-text pipeline that turned a massive volume of customer service calls into text, with AI analysis on top for operational and strategic decisions.
The problem
Customer service calls held a huge amount of information about what customers needed and where things went wrong, but as audio it couldn’t be analysed at scale.
What I did
I led the initiative as its product manager. Within my first six months we closed the sourcing process, defined the use cases, technical architecture and roadmap, put together a virtual AI development team, built the system integrations and had a working end-to-end solution.
The pipeline converted calls into text, and on top of it we built different AI solutions to analyse the data for different stakeholders, from day-to-day operations to strategic decisions. I owned the opportunity discovery, roadmap, backlog and prioritisation with the stakeholders.
Key technologies: NLP, speech-to-text and AWS.