I got tired of searching “cafe near me,” opening fourteen tabs, and somehow spending twenty minutes still not knowing where I wanted to go. Search gives options. It does not always give a decision.
So I built Hooligan: a mildly unhinged where-should-I-go-right-now agent. You can ask for a quiet cafe to work from within three miles, a fancy vegetarian lunch within walking distance of a client meeting, or an Indian restaurant near Venice Beach with a sunset view. The request is messy because real life is messy. That is the point.
Hooligan asks for location only when it actually needs it, discovers places that match the intent, considers the travel constraint, ranks the options, gives you a route, and lets you pick. I wanted the interaction to feel less like filling in filters and more like telling a friend what kind of day you are having.
The part I care about most happens after the visit. A little later, it checks in: did you actually like it? A yes becomes a favorite. A no is useful too, because the system should learn what not to recommend. That turns a one-off result into a loop: intent, context, discovery, action, feedback, memory.
Most local discovery products can answer what is popular near you. I am more interested in what is right for you right now, given what you are trying to do and what the product has learned over time. Hooligan is a small experiment in that bigger product question.