Answer
wavecat is a fully local personal agent that constantly watches your screen.
Ideally, it actively works in the background to understand your activity and develop a rich understanding of your needs and goals. It should be able to anticipate your needs, before you even ask.
Since all the models run entirely on your computer, all your data stays private and on your own computer. None of your personal data ever gets sent to the cloud.
Most agents wait for a prompt. But wavecat, by utilizing the on-screen context, also tries to offer suggestions for your next prompts so that you don’t have to even prompt as much. It grounds its suggestions in evidence from your screen, and this is shown to you so that you know that it knows what’s going on.
Even if you still prompt it and don’t use the suggestions, the help wavecat offers should hopefully be far more rich and grounded in your context. You can ask it things like “what was I doing yesterday?” and it’ll do a great job responding. No agent before wavecat could truly answer questions like that.
The more you use it, the more grounded its help becomes — see how it learns what you need.
There’s also nothing to configure after install, and it improves over time on its own.
The ultimate goal for the future is a genuinely useful personal agent that is private by default, because it lives right where your work already happens: on the computer you own, not in someone else’s data center.
It was developed by Samuel Yuan.