Open Innovation One PieceSubmitted July 18, 2026

CloseLoop AI - Meeting2Mission

An MCP app on the Model Context Protocol built by One Piece at the Amrita University Amritapuri campus NitroStack × MCP To The Moon hackathon and deployed on NitroStack.

About this project

Most meeting assistants stop at notes. CloseLoop AI continues after the meeting: it reads a transcript, extracts the real decisions (not just a summary), figures out who owns each one and by when, creates the Jira ticket, schedules the calendar deadline, posts the Slack follow-up, and later chases anything that goes overdue — with zero manual handoff between "we decided X" and "it's assigned, scheduled, and being followed up on." The Problem Meeting transcripts contain names and natural-language deadlines ("Sarah will finish the dashboard by Friday") — never email addresses, Jira project keys, Slack channel IDs, or ISO timestamps. Every meeting-assistant tool we tried either asked the user questions a transcript can't answer, or silently guessed and got it wrong. We treated that as a schema-design problem, not a prompting problem, and fixed it at the tool-contract level. How It Works Meeting → Transcript → Decision Extraction → Owner Resolution → Deadline Resolution → Jira Ticket → Calendar Deadline → Slack Follow-up → Escalation → Weekly Report (Notion + Email) → Executive Dashboard Real structured extraction via Gemini — decisions, owners, deadlines, priority, action items, risks, with a confidence score — not a vague summary. A human-in-the-loop confidence gate: anything under 80% confidence never becomes a ticket automatically; it's routed to a Slack approval request instead. An identity resolver that turns "Sarah Chen" into her real email/Slack/Jira/GitHub identity, backed by a small learned team-member directory, with an explicit ambiguous/not-found state instead of a silent wrong guess. A deadline resolver that turns "Friday", "next week", "end of sprint" into an ISO date relative to when the meeting actually happened. One-call orchestration (process_meeting) that runs the entire pipeline server-side instead of forcing the AI to manually chain 5+ tool calls.

Open Innovation track

Solve any real-world problem with AI, regardless of industry or domain.

Live MCP appSource code

Team One Piece

  • Sudhir Kumar SahLead

    X

  • Hridya Jain

  • V0R73X

  • Alish Karki

← All Amrita University Amritapuri campus projects

Frequently asked questions

What does CloseLoop AI - Meeting2Mission do?
Most meeting assistants stop at notes. CloseLoop AI continues after the meeting: it reads a transcript, extracts the real decisions (not just a summary), figures out who owns each one and by when, creates the Jira ticket, schedules the calendar deadline, posts the Slack follow-up, and later chases anything that goes overdue — with zero manual handoff between "we decided X" and "it's assigned, scheduled, and being followed up on." The Problem Meeting transcripts contain names and natural-language deadlines ("Sarah will finish the dashboard by Friday") — never email addresses, Jira project keys, Slack channel IDs, or ISO timestamps. Every meeting-assistant tool we tried either asked the user questions a transcript can't answer, or silently guessed and got it wrong. We treated that as a schema-design problem, not a prompting problem, and fixed it at the tool-contract level. How It Works Meeting → Transcript → Decision Extraction → Owner Resolution → Deadline Resolution → Jira Ticket → Calendar Deadline → Slack Follow-up → Escalation → Weekly Report (Notion + Email) → Executive Dashboard Real structured extraction via Gemini — decisions, owners, deadlines, priority, action items, risks, with a confidence score — not a vague summary. A human-in-the-loop confidence gate: anything under 80% confidence never becomes a ticket automatically; it's routed to a Slack approval request instead. An identity resolver that turns "Sarah Chen" into her real email/Slack/Jira/GitHub identity, backed by a small learned team-member directory, with an explicit ambiguous/not-found state instead of a silent wrong guess. A deadline resolver that turns "Friday", "next week", "end of sprint" into an ISO date relative to when the meeting actually happened. One-call orchestration (process_meeting) that runs the entire pipeline server-side instead of forcing the AI to manually chain 5+ tool calls.
Who built CloseLoop AI - Meeting2Mission?
CloseLoop AI - Meeting2Mission was built by team One Piece at the Amrita University Amritapuri campus NitroStack × MCP To The Moon hackathon, in the Open Innovation track.
What is an MCP app and how is it built?
An MCP app is an application built on the Model Context Protocol — an open standard that lets AI agents connect to tools, data, and APIs. This project exposes MCP tools and resources that agentic AI systems can call. It was built and deployed on NitroStack, the full-stack platform for shipping MCP apps and servers.