Reinventing customer service in the machine learning age

What it shows

Fourteen minutes, eight chapters, every screen real. The film follows one retailer from the usual stack of deflection tools to a service that answers, acts, learns and gets ahead of the next contact, and it shows the platform doing each of those things rather than describing them.

The argument runs in five beats. Answer everything, because the knowledge that customers need is already in the building and the job is to connect it. Act on everything, because an answer that cannot complete the task is only a better-worded deflection. Build it in days, because the people who own the service should be able to author and automate it without a bureau in the way. Learn from every conversation, because the contacts you handle are the clearest account you will ever get of what the service is really like. Then get ahead of the next one.

Why a film

Most writing about machine learning in customer service describes an outcome and asks you to take the mechanism on trust. This does the opposite. Everything on screen is the production platform, doing the work in front of you, so the claims can be checked against what you can see.

The platform in the film is Syntalyze, and the case study sets out the architecture underneath it: intent-driven orchestration, an authoring CMS and codeless workflows, Voice-of-Customer analytics, and the learning loop that turns each interaction into the next improvement. The thinking behind that loop is in Circles, not lines, and the reason the conversation is split between a model and a deterministic process is in One conversation, two engines.