Inspiration
Current Agents are unreliable because the prompt input space is too large. LLM Chatbots require effort for prompt formulation and task switching which forces you to stop focusing on your work. Adding other modalities (screenshots) can help with decreasing prompt length, but this adds more effort on the user to capture the screenshot.
Most current high-impact use cases for LLMs/Agents can be boiled down into a handful of simple repeatable tasks (fixed input prompts) along with current context from your desktop.
What if you could trigger reliable agents with zero effort? Almost like using hotkeys to control applications? Enter HotAgents!
What it does
Once the ctrl+space hotkey combo is pressed, HotAgents takes a screenshot of your desktop and feeds the screenshot into our agent (created with WordWare). This agent will then figure out which appropriate action to take based on the context of your screenshot, currently implemented actions include: -Content Explanation -Drafting a response message to chats/emails -Create code to implement/recreate what is on screen -Proofreading WIP documents
Log in or sign up for Devpost to join the conversation.