🧠 From Manual Apps to Intelligent Agents: Evolving How We Use Computers

Today, using a computer still means juggling apps, clicking buttons, and micromanaging tasks manually. But what if your computer could understand what you want — and act on it?

In this track, you’re invited to rethink the way we interact with technology: building intelligent agents that operate computers like humans do, but better.

Your mission is to design AI agents that see the screen, understand natural instructions, and handle real workflows — moving across apps, making decisions, and adapting on the fly.

That could mean (these are just examples - we expect you to come up with something of your own):

• An assistant that files your documents and cleans up your workspace after a project
• An agent that organises meetings, resolves scheduling conflicts, and follows up automatically
• A workflow partner that researches information, writes summaries, and updates your notes

These are just hints — there’s tons you can invent.

Focus on solving real digital work pain points and showing how a smart, perceptive AI could collaborate with humans across everyday computer tasks.

🎯 Your Mission

In 24 hours, your team will design and prototype an AI agent that operates a computer through the GUI to solve a practical productivity problem.

You can build on top of existing vision-language frameworks — or combine APIs, GUI automation tools, and your own logic to craft a working demo.

We’re not looking for a basic bot.

We want to see an AI agent that:

• Understands natural human instructions (typed or spoken)
• Perceives the screen, apps, and UI dynamically
• Makes decisions based on context
• Coordinates multi-step workflows across multiple apps