Zero | Devpost

Inspiration

Inspired by Sci-fi, we imagine humans to be one with technology. With zero restrictions, zero accessibility barriers. We see this as the foundation for the next generation of human computer interfaces coming in the intelligence era. Building accessibility first, we try to echo what BCIs are meant for. All the utilities, at some point, in the future would be abstracted away, we will only have screens for entertainment.

What it does

We built a copilot which becomes one with you, interact with the internet seamlessly with voice commands in any language, having zero barriers to anything. Order a coffee, Write an email, absolutely anything that you can think of.

How we built it

We think of it as using LLMs as an assistant, for each task that the user prompts, we parse the Document Object Model for each page we land on, query the LLM to figure which action to take and then we take the action.

Challenges we ran into

One of the biggest challenges we ran into was narrowing the vision down, building it niche first and then scaling from there.

Technically another challenge, optimizing the DOM to be smaller as to not hit the token limits on the LLMs, Building some usecases.

Accomplishments that we're proud of

This project was inspired by another model architecture called Large Action Models, a Model that operates on Actions. Us managing to create the same use with large language models, Using Large Language models for the same use as Large Action models, without the LAM bottlenecks was one achievement.

What we learned

We learned about accessibility and how computers are still not acessible as much as we'd think.

What's next for Zero

We want to ultimately build a really powerful large action model that serves as an operating system for the next generation of Human Machine Interfaces, this project was a proof of concept for what can be possible. In the future, we imagine this to be used by companies like wispr as the OS for their BCI and we also imagine this to be the accessibility first ubiquitous way to interact with computers of the future.