Case Study: Spoke boosts AI-powered workplace ticketing productivity and eliminates weekly DB firefighting with MongoDB Atlas

Building an AI-powered workplace ticketing system on MongoDB Atlas at Spoke

Spoke is an AI-powered workplace support platform that automates answers to routine employee questions across Slack, email, SMS, and web by building custom models per customer. The team struggled with a complex, evolving data model (knowledge articles, ephemeral chat sessions, callback maps) and scaling limitations with their previous Compose.io MongoDB hosting, which led to painful debugging episodes and a lack of operational visibility.

Spoke migrated to MongoDB Atlas in 15 minutes using Live Migration and now uses MongoDB’s flexible document model and indexes to store models, articles, and ChatSession state. The move unlocked far better observability (from 3 to nearly 30 metrics), built-in alerts and optimization suggestions, eliminated recurring weekly troubleshooting, and freed engineers to focus on product roadmap features like location-aware personalization.

Spoke

Nate Rook

Senior Software Engineer