Inspiration
After 5 years as a data engineer, I noticed most enterprise AI projects stall after impressive demos — because 80% of the real work lies in connecting and preparing data. This data plumbing is the biggest bottleneck between prototype and production.
What It Does
Twig automates private data integration for enterprise AI. With 30+ secure, enterprise-grade connectors, it handles ingestion, OAuth, chunking, and vectorization automatically. Engineers can simply add the Twig Private Data MCP Server as a tool to their AI stack and start shipping faster.
How We Built It
Twig consists of two core layers:
- Data Layer: Python, Prefect, NLTK, Pandas, AWS Fargate/ECS clusters.
- MCP + Frontend: Built with Next.js on Vercel, using Redis for caching and orchestration.
Challenges
Integrating authentication with the MCP protocol was tricky — especially since MCP tools currently lean heavily on OAuth. We’re exploring better ways to support API token–based access.
Accomplishments
Twig’s MCP Server now works seamlessly with Claude MCP — enabling enterprises to connect private data instantly and use it in production AI workflows today.
What We Learned
Automating the data layer can cut enterprise AI development time by up to 80%, turning demos into deployable systems.
What's Next
Harden, scale, and commercialize the Twig Private Data MCP Server.
Log in or sign up for Devpost to join the conversation.