Create an account to save sources and durable progress.
Product Talk · Teresa Torres
articleplaybookSource excerpt
About Product TalkListen to this episode on: Spotify | Apple PodcastsWhat does it take to build an AI voice agent that people will talk to for 44 minutes about their own blind spots? Your360 AI found out, one call at a time.In this episode of Just Now Possible, Teresa Torres talks
This is a limited feed-provided excerpt, not the full original work.
Apollo layer
Build complex AI experiences as controlled systems, not single prompts: use state machines to manage the interaction, evals to measure quality, and layered synthesis to turn long conversations into useful outputs. Improve the system call by call.
For sensitive, extended interactions such as feedback conversations, a compelling interface is not enough. Founders need a repeatable way to control behavior, assess results, and refine quality as real usage reveals failure modes.
Map one critical AI workflow into explicit states, then define an eval for each transition and a layered process for producing the final output. What can you learn from the next 10 real interactions?
Private notes
© Apollo Advisors° ABN 34 346 108 139