Momentum

An autonomous AI workforce for your desktop

Comment

Gemini Momentum

Gemini Integration Description

Momentum leverages Gemini 3's advanced capabilities to power autonomous AI agents for email and file management.

Function Calling & Tool Use: The application defines 15+ tools for file operations, Gmail search, Google Sheets export, and storage analysis. Gemini 3 Flash intelligently chains these tools to complete complex multi-step tasks from single natural language commands—like "pull all Instacart receipts from Gmail and create an expense spreadsheet."

Gemini Vision: Receipt images and documents are analyzed using Gemini's multimodal understanding. The system extracts structured data (vendor, date, amount, individual line items) from grocery receipts, invoices, and photos—enabling automatic expense categorization and smart file renaming based on visual content.

Autonomous Email Agents (Orbits): Up to 5 Email Watchers continuously poll Gmail, using Gemini to classify emails by intent rather than keywords. A rule like "find Instacart receipts" correctly identifies order confirmations while ignoring marketing emails. Matched emails trigger automated workflows: data extraction, Google Sheets logging, Gmail labeling, and archiving.

Autonomous File Agents: File Watchers monitor local folders, automatically organizing and processing files using Vision analysis.

2-Layer Orchestrator: A routing layer classifies task complexity, optimizing cost by directing simple operations to Flash-minimal and complex reasoning to Flash-high.

Full Project Description

Momentum - Your Autonomous AI Workforce

The Problem

The average knowledge worker spends 2+ hours daily managing email and files. Receipts get buried in Gmail. Downloads folders become chaos. Expense tracking requires tedious manual data entry. What if AI could handle all of this automatically?

The Solution

Momentum is a cross-platform desktop application that combines intelligent chat assistance with autonomous AI agents called "Orbits." Unlike chatbots that wait for commands, Momentum's agents work continuously in the background—monitoring your email, organizing your files, and extracting data while you focus on what matters.

Key Features

💬 Intelligent Chat Mode

  • Natural language file operations and document analysis
  • Storage analyzer with interactive charts and cleanup recommendations
  • Full Google integration: search Gmail, download attachments, export to Sheets
  • Delete reviewer for safe, reversible file operations

📧 Email Watchers (Orbits)

  • Autonomous Gmail monitoring with configurable polling intervals
  • AI-powered classification by intent, not keywords
  • Automatic actions: extract data to Sheets, apply labels, star, archive, mark read
  • Example: "Watch for Instacart receipts → log items to expense tracker"

📁 File Watchers (Orbits)

  • Monitor folders for new files automatically
  • Gemini Vision extracts data from receipt images
  • Smart rename: IMG_001.jpg → 2026-01-17_Starbucks_$8.50.jpg
  • Automatic categorization and organization

🔒 Safety First

  • Delete review panel—approve before any files are removed
  • Trash system with full restore capability
  • Runs in system tray, agents continue while minimized

What Makes It Different

Feature Momentum Traditional Tools
Operation Autonomous (24/7) Manual (on-demand)
Understanding AI comprehends intent Keyword matching
Platform Windows, Mac, Linux Often single platform
Cost Free Gemini API tier Expensive subscriptions

Demo Use Case: Grocery Expense Tracking

One command: "Search Gmail for Instacart receipts from the last 2 months and create an expense spreadsheet"

Momentum searches Gmail, downloads receipt attachments, uses Gemini Vision to extract every item with prices, categorizes spending (produce, dairy, snacks, etc.), and creates a formatted Google Sheet—all automatically.

Then set up an Email Watcher, and every future Instacart order gets logged automatically. Forever.

Third-Party Integrations

APIs & Services:

  • Google Generative AI (Gemini 3 Flash) - @google/generative-ai v0.24.1
  • Google APIs (Gmail, Sheets, Drive, OAuth) - googleapis v170.1.0

Document Parsing Libraries:

  • mammoth v1.11.0 - DOCX parsing
  • pdf-parse v2.4.5 - PDF parsing
  • xlsx v0.18.5 - Excel/spreadsheet parsing
  • papaparse v5.5.3 - CSV parsing

Core Framework & Libraries:

  • Electron v39.2.6 - Desktop application framework
  • React v19.2.1 - UI framework
  • TypeScript - Type safety
  • Tailwind CSS v4.1.18 - Styling
  • Zustand v5.0.10 - State management
  • Chokidar v5.0.0 - File system watching
  • Recharts v3.6.0 - Data visualization
  • Framer Motion v12.33.0 - Animations
  • Electron Store v8.2.0 - Local configuration storage

All libraries used under their respective open-source licenses (MIT, Apache 2.0).

Built With

Share this project:

Updates