Translate possibility into a product decision.
I explore what new models can do, learn directly from customers, and use both to shape the product experience. The interesting part is finding where technical progress makes a real customer problem easier to solve.
Choose a model for the conversation it enables.
Consider a caller who interrupts an appointment request to change the day. I’d compare candidate voice models using the same scenario: how quickly each responds, whether it stops speaking when interrupted, and whether it carries the corrected request through to the next step. I’d weigh those results alongside cost per completed task and handoff behavior. A natural-sounding voice is only useful if the conversation stays responsive and the caller can finish what they came to do. This illustrates how I’d evaluate the tradeoff; it is not a reported CallJoey launch or experiment result.
Keep product and market in the same conversation.
My work connects product decisions with how an experience is understood and brought to market. ExcelPal gave me a practical feel for messaging and distribution; working in AI brings that perspective into a changing technical environment.
A capability matters when it becomes useful to someone.