Recall Ray - AI Tinkerers - Nashville Hackathon
AI Tinkerers - Nashville
Hackathon Showcase Finalist

Recall Ray

Team led by John Berryman, Principal Consultant at Arcturus Labs — senior AI/ML engineer (Virginia Tech), 10+ years, LLMs, RAG, FastAPI/Flask, React, OpenAI/Anthropic.

3 members Watch Demo

Recall Ray is a centralized portal powered by the Model Context Protocol (MCP) that connects Gmail and Google Calendar to automatically generate daily agenda cards, surface high-risk items, and provide a context interface where users can read and write emails or update calendars without leaving the platform. Knowledge workers spend nearly 20% of their week searching for information, while managers lose close to 30% switching between tools and often miss critical, high-risk tasks. Recall Ray addresses this by consolidating context and execution in one workspace, ensuring important details aren’t overlooked.

During the hackathon, we built and shipped the core functionality: Agenda, High-Risk, and Context Interfaces that run on live Gmail and Calendar integrations. Feedback from managers and knowledge workers validated the need for a single view that highlights risks and cuts context-switching time. The UI is card-based, lightweight, and intuitive, designed to match existing workflows. On the backend, a Node.js MCP orchestrator standardizes tool calls, while the React + Tailwind frontend provides a clean experience.

The foundation is MCP-first, making it easy to extend: Slack and Google Drive integrations are next, enabling features like chat context retrieval and PDF summarization. We also have tool-call logging, lineage, and latency metrics on the roadmap to enhance transparency and enterprise readiness. By reducing wasted time, preventing missed high-risk items, and unifying action with context, Recall Ray demonstrates strong functionality, user validation, design, feasibility, and execution — and it is ready to evolve into a trusted, enterprise-scale productivity platform.

Targeting tim conners productivity challenge.

We had been planning the architecture and engineering approach for the past four weeks, focusing on how to structure a multi-tool, MCP-first system that could scale to additional integrations. This preparation included conceptual design, technical research, and outlining the connector model for Gmail, Calendar, Slack, and Drive.

All of the actual implementation — including the Agenda card, High-Risk surfacing, Context Interface, Gmail and Calendar integrations, and the working React + Node.js codebase — was built during the hackathon itself. The prior work gave us a clear blueprint, but the functioning product you see now was created entirely within the hackathon timeframe.

AI Tinkerers APIs: Gmail API Auth.js (auth) Backend: Node.js Collaboration & Productivity: GitHub (repo & version control) Database: MongoDB (session/user data) Express.js Figma (UI design mockups) Frontend: React Google Calendar API Libraries & Utilities: fetch(API calls) Other Helpful Tools: Google Workspace test accounts for integration validation Postman (API testing) Protocols: Model Context Protocol (MCP) for standardized tool calls Slack (team coordination during hackathon) Tailwind CSS (UI) The Lighthouse dotenv (config)

The main project

Summarizing URL...

Loom video submission if the youtube link doesn't store the link