How to Manage Your AI Agents Directly in Slack

Quick Overview

Mighty Stack Workflows allow users to trigger multi-agent processes directly from Slack messages, demonstrated by a workflow that triggers on a Slack message, uses a Research Agent to gather data (like company office locations via Google Search), and then passes the findings to a Slack Updater Agent to format and post a detailed reply in a thread using Block Kit for rich formatting.

Key Points: The demonstrated workflow triggers when a user message is received in Slack, allowing for multi-agent collaboration directly within the platform. The workflow employs a Research Agent utilizing the Google Search tool to find specific information, such as OpenAI's office locations, which was found to be complex/unavailable in Sydney but present in other locations like San Francisco, London, and Dublin. The Research Agent passes its findings to a Slack Updater Agent, which formats the output using Slack's Block Kit for rich formatting, including bolding, italics, and user tagging. The Slack Updater Agent uses the advanced 'Send Message - Slack Workforce' tool to structure the output, including user mentions (like tagging Josh Jackson) and attaching images (like the OpenAI logo). The video shows the final output in a Slack thread, confirming the status as 'Completed' and displaying the well-formatted, researched information. Users can optionally configure the Slack trigger to listen only to specific channels, direct messages, or a combination thereof, offering granular control over automation activation.

Context: The video demonstrates a new feature in Mighty Stack Workflows that integrates AI agents directly with Slack, allowing workflows to be initiated by incoming Slack messages. The presenter walks through building a specific workflow designed to answer sales-related questions about a company (OpenAI) by using a multi-step process involving research and updating Slack with formatted responses.

Detailed Analysis

The video details how to set up an AI agent workflow in Mighty Stack that triggers directly from Slack messages, enabling automated research and response generation. The core structure involves a Slack trigger, followed by a Research Agent that executes a Google Search tool to gather specific data—in this case, OpenAI's office locations. The research results, including complex findings like the lack of a Sydney office, are then passed to a Slack Updater Agent. This final agent uses the advanced 'Send Message - Slack Workforce' tool, which supports Slack Block Kit formatting, to generate a highly structured, formatted reply containing text styling, user tags (like tagging Josh Jackson), and even embedded images (like the OpenAI logo), all delivered within a Slack thread. The presenter highlights that the Slack trigger supports specifying keywords, excluded keywords, and listening in specific channels or direct messages. Upon running the workflow, the status quickly updates to 'Completed' in Slack, and the resulting message thread displays the fully formatted, researched content, demonstrating powerful, automated lead qualification or information retrieval directly within the communication platform.