How We Built Brian: Our Custom AI Agent for Slack

Published October 15, 2026

Learn how we built Brian, our custom Slack AI coworker designed to automate daily tasks, manage workflows, and connect information across tools like GitHub, Linear, and Notion. In this walkthrough with Aarish, we explore Brian's architecture, how it uses AI agents, MCP integrations, memory, and sandbox environments to execute tasks, and why we chose to build our own workspace agent. We also discuss when to use Slack agents versus local developer tools like Cursor and Claude Code, the tradeoffs of building versus buying, and what teams need before implementing their own AI coworkers. ⏱ Chapters: 0:00 Introduction 1:19 Why We Built Brian 4:06 Slack Agents vs IDE 5:41 Agent Architecture 11:36 Setup and Configuration 13:24 Custom Agents 15:15 Build vs Buy 18:56 Common Problems and Namespaces 20:53 Prerequisites and Advice #AIAgents #SlackAI #AgenticAI #WorkflowAutomation #MCP #AIEngineering #SlackAutomation #Brainforge ============================= Brainforge helps teams build scalable data platforms and production AI agents that turn complex data into clear, actionable insights. ⚡ Learn more about what we're building: 👉 https://www.brainforge.ai/ 📅 Join upcoming Brainforge events: 👉 https://luma.com/brainforge ============================= Connect with Us LinkedIn: https://www.linkedin.com/in/ukumaran/ LinkedIn: https://www.linkedin.com/in/tsengr/