About Acta
Meetings are where decisions happen, but rarely where execution happens. Teams spend hours discussing ideas, yet the follow-up work often gets lost across notes, emails, calendars, and task managers.
We built Acta to close this gap.
Acta is a meeting execution engine that turns meeting transcripts into structured summaries, key decisions, and actionable tasks. Users can review results, approve actions, and execute them directly into real workflows like Google Calendar.
Instead of just summarizing conversations, Acta focuses on what should happen next.
conversation → structure → execution
Inspiration
We were inspired by how much time teams waste after meetings trying to remember what was decided and who is responsible for what. Most tools stop at transcription or summarization, but the real problem is turning discussion into action.
We wanted to build something that helps teams move faster and with more clarity after meetings.
How we built it
We built Acta using FastAPI, SQLite, and SQLAlchemy for the backend, with Google Gemini (GenAI) to extract structured summaries, decisions, and action items from transcripts.
The frontend is built with HTML, CSS, and JavaScript, providing a clean interface where users can review, approve, and execute tasks. Approved actions can be pushed to tools like Google Calendar, demonstrating real workflow integration.
Challenges we faced
One challenge was handling ambiguity in transcripts, since tasks are often implied but not clearly assigned. We added confidence scoring and review states to ensure users stay in control.
We also worked through frontend-backend integration and version control issues involving large media files exceeding GitHub limits.
What we learned
We learned that AI tools need both automation and human review to be useful. Building structured outputs that users can trust is just as important as generating them.
Log in or sign up for Devpost to join the conversation.