The evolution from “AI second brain” toward a broader AI operating system is a pretty significant shift in scope. Curious how Hatch has progressed since this post — has the product ended up validating that broader vision, or has usage pulled you in a different direction?
The "window as the primitive instead of chat" framing is a genuinely different take, most AI assistant products just wrap a chatbot in nicer UI and call it a second brain. I'm building in the adjacent space with FounderFlow (mentioning since it's relevant), more narrowly focused on daily triage than a full AIOS, so I've been wrestling with a smaller version of the same problem: how much should live inside the assistant's memory vs. just get referenced from where it already lives. Have you landed on a rule for when something graduates from "referenced context" to something Hatch actually stores and reasons over?
Hi, I'm very new to all this and just about to launch a SaaS app in the accountability /PT space, so sort of related but not directly. That research has got me more interested in AI Assistants and where that is heading, particularly with the arrival of Tomo, but also with the precedent / vision of Kubrick's HAL. Bit of a side question to your product, but how close do you think you could get Hatch to operate as HAL ?
About
We're building Hatch because the next generation of AI assistants should help you do real work, not just answer one-off questions in disposable chats.
3 Comments
The evolution from “AI second brain” toward a broader AI operating system is a pretty significant shift in scope. Curious how Hatch has progressed since this post — has the product ended up validating that broader vision, or has usage pulled you in a different direction?
The "window as the primitive instead of chat" framing is a genuinely different take, most AI assistant products just wrap a chatbot in nicer UI and call it a second brain. I'm building in the adjacent space with FounderFlow (mentioning since it's relevant), more narrowly focused on daily triage than a full AIOS, so I've been wrestling with a smaller version of the same problem: how much should live inside the assistant's memory vs. just get referenced from where it already lives. Have you landed on a rule for when something graduates from "referenced context" to something Hatch actually stores and reasons over?
Hi, I'm very new to all this and just about to launch a SaaS app in the accountability /PT space, so sort of related but not directly. That research has got me more interested in AI Assistants and where that is heading, particularly with the arrival of Tomo, but also with the precedent / vision of Kubrick's HAL. Bit of a side question to your product, but how close do you think you could get Hatch to operate as HAL ?