There are a lot of startups that are treating artificial intelligence as a product to help answer questions or generate artifacts like text or images for specific purposes. But there aren't many tools that help people make complex decisions or accomplish tasks in a way that scales.
At the founding of the field, artificial intelligence was thought to consist of machines that could perceive their environment and use learning and reasoning to take actions that help achieve defined goals.
So why not try to build that?
With modern software, especially with generative AI, we have the ability to create truly reactive, dynamic digital experiences that respond to user needs in a way that software even half a decade ago didn't do. Our systems aren't perfect, but they're useful enough to build on. Yet right now, we're focused on reinventing the wheel by building brittle chatbots that don't solve problems for many people.
We want to explore the possibilities of AI as a tool for decision-making and genuinely getting things done. Can we create a reliable system that uses generative interfaces to collect and display information based on arbitrary context? Can we create intelligent, stateful systems to automate the most tedious of tasks? Can we create a reasoning platform that is flexible enough to help users accomplish the most complex of goals?
While it's likely that the current trend of generative AI tools is a bubble, that doesn't mean it hasn't created entirely new interaction patterns that can genuinely extend what's capable through software. As such, right now is the perfect moment to begin creating the next generation of software tools that use AI as less of an engine and more of a glue that binds together the user's intent with the systems that can help them achieve their goals.
(Project) Haystack aims to create a platform for planning, making, and executing decisions. We're going to push the boundaries of what it means to build intelligent systems that help people get things done.