Cortex: LLM-Powered Agentic OS

A walkthrough of Cortex, an LLM-driven OS, covering Explorer file management, Deep Research document‑internet integration, and Maestro financial workflow automation using LangGraph and Gemini API.

Overview

I’ll present a technical deep-dive into Cortex – Agentic OS, an LLM-powered operating system designed to manage modular digital knowledge workflows. The demo covers three specialized applications:
• Explorer: Technical walkthrough of the file-management system, demonstrating how documents and generated outputs are uploaded, organized, and tracked through AI-enhanced workflows.
• Deep Research: Behind-the-scenes look at how this module combines user-uploaded documents with live internet data, showing internal document search implementation, Gemini-powered content summarization, Google search integration, and the logic for automated report grading.
• Maestro: Detailed code-level exploration into Maestro’s integration with financial databases, demonstrating automated workflow creation for competitive analyses and annual reports, and showcasing specialized tooling for competitor analysis, business highlight extraction, and detailed company description analysis.

Expect detailed code walkthroughs, live examples, and practical insights on agent orchestration using LangGraph and Gemini API, emphasizing the engineering challenges and solutions behind Cortex’s intelligent modular design.

Tech stack

Related projects

AI-Powered Terminal Workflows: Integrating Gemini CLI with Jira MCP

Montreal

Demonstrate how to connect Gemini CLI with Jira's Model Context Protocol, enabling command‑line creation, planning, and execution of…

Setting up a chatbot with Google Gemini

London

Demo building a Gemini Pro chatbot using Streamlit, LangChain, and Vertex AI in under 100 lines, focusing on…

Learn how to Augment a Vertex Gemini application with open source predictive ML models from Hugging Face.

Seattle

Learn to deploy Hugging Face TimesFM via Vertex Model Garden and integrate it with a Vertex Gemini retail…

Using Gemini's Large Context and Schema-Enforced Function Calling to Generate Production Kubernetes & Sveltos YAML Configurations

San Francisco

Explore how Gemini’s large‑context and function‑calling generate production‑grade Kubernetes and Sveltos YAML, using a multi‑stage agent and schema‑enforced…

Signal from the Noise: Using Gemini AI to find great Jazz

Boston

Learn how to use Gemini's API to extract structured artist and track data from jazz blogs, summarize articles,…

Generative AI on Google Cloud: Tools, Models, and Real-World Applications

Vancouver

Explore deploying Gemini multimodal models on Google Cloud with Agent Builder, Vertex AI, and Vertex Search, featuring demos…