Inspiration
We built Thread.Ai after noticing how easy it is to forget important details after meetings. During internship calls, info sessions, office hours, or work meetings, people are trying to listen while also remembering deadlines, links, follow-ups, and next steps. Even when notes are taken, the actions hidden inside those notes often never get completed. We wanted to answer one question: what if a meeting could help complete its own follow-up?
What it does
Thread.Ai listens to meetings in real time and turns important moments into actions. It identifies opportunities, deadlines, resources, requirements, events, decisions, and follow-up tasks. Instead of only creating notes or summaries, Thread.Ai can help fill out forms, draft emails, create calendar events, and detect links or QR codes shown during screen sharing. Before anything is sent or submitted, the user reviews and approves it.
How we built it
We built Thread.Ai as a web application and meeting side panel that works alongside platforms like Zoom and Google Meet. Real-time transcription captures the conversation, while AI analyzes the transcript to identify important moments and possible actions. We also connected the system to Gmail and Google Calendar so approved actions can happen directly. For shared screens, Thread.Ai can analyze visual content and recognize resources such as QR codes and links.
Challenges we ran into
One of our biggest challenges was making sure the AI could take useful actions without making up information. For example, when filling out an application, every answer needed to come from the meeting or the user's provided information. We also had to handle real-time transcription, screen-share information, third-party integrations, and user approvals while keeping the experience simple and trustworthy.
Accomplishments that we're proud of
We successfully demonstrated the entire workflow from conversation to completed action. Thread.Ai filled out a 23-field internship application using information from the meeting and the user's resume, sent real follow-up emails through Gmail, created a real Google Calendar event, and detected a QR code displayed during a screen share. Most importantly, every external action required the user's approval before it happened.
What we learned
We learned that meeting summaries are useful, but they are only part of the problem. People often already know what happened in a meeting — the harder part is actually completing everything that needs to happen afterward. We also learned that AI agents become much more useful when users stay in control, especially when the AI can send emails, submit forms, or interact with personal accounts.
What's next for Thread.Ai
Next, we want to make Thread.Ai work across more meeting platforms and support more types of actions and integrations. We also want to improve how it understands context across multiple meetings so it can help users keep track of opportunities, deadlines, and ongoing conversations over time. Our goal is to make Thread.Ai a trusted assistant that turns every important conversation into completed action — don't just take notes, take action.
Log in or sign up for Devpost to join the conversation.