Product and engineering leader who has built large-scale data and AI systems at companies like Toyota and Netflix. Writes about the architecture behind conversational analytics.
Most companies measure an AI agent's cost from the vendor bill. In a typical servicing example, that bill is less than a fifth of the true cost; the rest sits in reviews, hand-offs to people and repeat work.
You cannot trust an AI agent you cannot measure, and you cannot measure one until you can read the KPIs it is meant to move. That takes a model of your business first.
Conversation as Visualization is our vision for the next era of understanding: a continuous exchange between human storytelling and machine visualization, where data questions trigger visual responses and visuals spark new questions.