STAR

AI / LLM CHATBOT

During my time at Telefónica, I was brought in to lead a discovery workshop addressing a persistent pain point: customers were experiencing long wait times and being misrouted when trying to resolve everyday queries. Over time, this was creating a noticeable breakdown in trust between customers and the business — and needed a solution grounded in real insight rather than assumptions.

SITUATION

Customers were being misrouted or waiting too long to reach the right department, creating friction and eroding trust at scale.

TASK

Design and deliver an AI-powered chatbot using large language models that could accurately route customers to the right department — covering both standard journeys and complex edge cases

ACTION

Conducted in-depth user interviews and admin team interviews to surface real customer needs and pain points
Listened back to customer call recordings to identify patterns and recurring query types
Mapped both happy path journeys and a full range of negative/edge case scenarios to ensure comprehensive coverage
Used insights to train and refine the LLM-based chatbot, grounding it in real customer language and intent

RESULT

The chatbot significantly reduced misrouting and wait times, improving first-contact resolution and customer satisfaction — built entirely on genuine insight rather than assumptions.