NaviAI

• 

  Navi Logo

Inspiration

We developed NaviAI with three core goals in mind: expanding device interaction options, ensuring universal ease of use, and setting a new standard for intuitive technology engagement. Our inspiration stemmed from the desire for a simpler and more accessible browsing experience.

What it does

NaviAI converts your verbal commands into browser actions. You say something like "Click the submit button", and navi ai will click it for you

How we built it

We build NaviAI ai through a mixture of LLMs, function calling, structured LLM output, and determination :)

Challenges we ran into

• LLM hallucinations
• 20 images max from Claude
• Push-to-talk microphones
• Not exiting early in JS closures

Accomplishments that we're proud of

Getting an end-to-end browser action system that can respond to verbal commands, that's pretty accurate!!

What we learned

Teamwork! No seriously, we all collaborated and joined our unique skills together.

What's next for NaviAI

We are going to ship a Chrome extension with all of our browser action logic for a seamless experience.

Built With

• anthropic
• chrome
• claude
• gpt
• media-recorder
• nodej
• openapi
• typescript