Atlas — Mapping How Organizations Really Work

Inspiration

During my internships as an IT Infrastructure Analyst at Textron and a Software Developer at Gainwell Technologies, I realised that many of the biggest engineering challenges weren't caused by technology— they were caused by how teams shared knowledge.

I often saw important knowledge concentrated in one or two experienced team members, documentation that was unnecessarily long, complex, or outdated, approvals that delayed implementations and releases, and new team members struggling to understand the team's workflows before they could contribute independently. At the same time, Slack conversations grew quickly and became cluttered, making it difficult to find important decisions or understand the context behind them.

These experiences inspired me to build Atlas—an AI-powered organisational intelligence agent that transforms everyday Slack conversations into actionable insights, helping teams identify hidden risks, knowledge bottlenecks, workflow inefficiencies, and documentation gaps before they become larger operational problems.

What it does

Atlas is an AI organisational intelligence agent built for Slack that helps teams understand how work actually happens.

Instead of acting like a traditional chatbot, Atlas analyses Slack conversations to identify knowledge silos, documentation gaps, workflow bottlenecks, approval delays, onboarding challenges, and collaboration patterns. It then generates executive-ready reports, dependency maps, action plans, and trend analysis to help engineering managers and team leaders make faster, more informed decisions.

How I built it

Atlas is built using Slack Bolt for JavaScript, the Slack Agent Builder, the OpenAI Agents SDK, and the Slack Web API.

Atlas retrieves recent Slack messages from a channel, analyzes communication patterns with AI, and generates structured organizational intelligence directly inside Slack threads. I also built a Slack App Home dashboard, reusable prompt workflows, command routing, and multiple reporting capabilities including Organizational Health Reports, Risk Radar, Dependency Maps, Operational Action Plans, Executive Briefs, Scorecards, and Trend Analysis.

Challenges I ran into

One of the biggest challenges was teaching AI to identify meaningful organizational patterns instead of simply summarizing conversations.

I spent a significant amount of time refining prompts so Atlas could consistently detect knowledge concentration, workflow bottlenecks, documentation debt, onboarding friction, and operational risks while avoiding unsupported conclusions. I also learned how to work with Slack's permissions, event handling, API integrations, and securely manage OpenAI API keys throughout development.

Accomplishments that we're proud of

I'm proud that Atlas goes beyond being another AI chatbot.

Instead of simply answering questions, Atlas transforms everyday Slack conversations into practical organizational intelligence that leaders can immediately act on. I'm especially proud of building a complete workflow that includes Organizational Health Reports, Risk Radar, Dependency Maps, Executive Briefs, Operational Action Plans, Trend Analysis, and a Slack-native App Home dashboard—all designed to help teams proactively improve how they work.

What we learned

Building Atlas taught me that the hardest part of AI isn't generating responses—it's identifying the right problems to solve.

I learned how to design Slack-native AI experiences, build with the OpenAI Agents SDK, structure prompts for consistent organizational analysis, and translate unstructured conversations into insights that are useful for both engineers and leadership. Most importantly, I learned that successful AI products solve real workplace problems instead of showcasing AI for its own sake.

What's next for Atlas — Mapping How Organizations Really Work

My long-term vision is for Atlas to become a comprehensive organizational intelligence platform that continuously helps teams understand how work flows across an entire Slack workspace.

Future plans include workspace-wide organizational health dashboards, historical trend tracking, cross-channel dependency analysis, real-time organizational risk alerts, deeper Slack MCP and Real-Time Search integration, workflow automation recommendations, and executive dashboards for engineering and operations leaders.

Ultimately, I want Atlas to help organizations discover hidden operational risks early, distribute knowledge more effectively, and build healthier, more resilient teams before small issues become major business problems.

Built With

• gpt-4.1-mini
• javascript
• node.js
• openai-agents-sdk
• slack-agent-builder
• slack-bolt-for-javascript
• slack-web-api