Case study · CustomerView
Thousands of calls, one at a time.
CustomerView scored call-center conversations with machine learning. Supervisors needed to move fast across thousands of calls; agents wanted to review their own, one at a time. Same product, very different speeds.
The heart of it was a call explorer: the audio as a two-track waveform so you could see who was talking, a transcript synced to it with each line marked positive or negative, and the agent’s scorecard alongside, compared against team benchmarks.
The filters were their own puzzle. Seven dimensions, from location to team to queue, could stack up to 40-plus active constraints, and a supervisor still had to see at a glance what was in and what was out, and clear one without losing the rest.
CustomerView was acquired partway through the build. The buyer wanted the data pipeline and shelved the interface, but the design was finished, tested, and already turned into working front-end components.
What are we making?
or email hello@schwaduct.com