AI Agent builders: how do you actually keep track what competitors are shipping, and figure out what product features to build next in crowded verticals like healthcare, finance, legal, etc?

Curious how people handle a few things, since I keep hitting the same wall myself. If you're building an AI agent in a space with real competition (not the only player): How do you find out when a competitor ships a new feature? Do you check manually, hear about it from a customer, or just not track it at all? When there are multiple use cases in your category, do you know which one you're actually winning, or losing, in? Or is that mostly a guess? How do you decide what product feature(s) to build next - customer requests, watching competitors, gut feel, something else? Do you have any real sense of what buyers are actually looking for before they talk to you like what they compare, what makes them pick one agent over another - or is that mostly a black box until they're already mid-conversation with you? Just curious whether this is something people actively track, or if it's one of those things everyone means to do and nobody actually does.