Inspiration

Managing email daily is hard. Most AI tools either force you to use a new inbox, install extensions, or worse, store your private data. I wanted a solution that respects privacy, integrates natively with your existing inbox, and saves hours every week.

What it does

  • Summarises long emails into bite-sized drafts
  • Writes intelligent replies directly into your inbox’s draft folder
  • Categorises emails into custom tags like work, payments, urgent, etc
  • Respects your privacy, no data is stored, and users can exclude specific emails from tracking
  • Works without extensions or platform changes, use your inbox as usual.

How we built it

Backend: Python with secure google auth access for inbox read/draft write AI Agent: Built using OpenAI’s GPT model with local context handling Email Integration: Used Gmail API for fetching, summarising, and writing drafts Security: No email data is stored; processing is stateless and ephemeral using AWS Lambda and Event Bridge Front-End: React Js

Challenges we ran into

  • Ensuring summaries and replies are context-aware without violating user privacy
  • Google auth compliances are complex and hard to work.
  • Handling varied formats and email structures cleanly and reliably.
  • Avoiding secure and unnecessary emails such as OTP, login creds, etc

Accomplishments that we're proud of

  • Fully functional, extension-less inbox assistant
  • Privacy-first design with zero email data storage
  • Seamless user experience, works natively in Gmail without changing user behaviour
  • High accuracy in auto-replies and email categorisation

What we learned

  • Working with Google complex Auth.
  • Handling real-world emails is messier than expected.
  • Security and privacy aren't just features.
  • Summarisation is more impactful.

What's next for InBo

  • Support for other email inboxes such as Outlook, ProtonMail, and other inbox providers
  • Team/Workspace mode to manage shared inboxes (e.g., support@yourcompany.com)
  • Add more smart actions like reminders, follow-ups, and calendar event suggestions

Built With

Share this project:

Updates