Cloudflare Launches Town Lake and Skipper: AI-Powered Data Platform for Unified Analytics
Key Takeaways
- ▸Cloudflare built Town Lake and Skipper to eliminate data fragmentation affecting 1B+ events processed per second
- ▸Town Lake provides a single SQL interface with automatic PII detection and time-bounded, auditable access controls
- ▸Skipper enables non-technical employees to ask data questions in plain English and receive accurate answers in seconds
Summary
Cloudflare has announced two new internal tools designed to solve a critical data infrastructure challenge: Town Lake, a unified SQL data analytics platform, and Skipper, an AI agent that enables employees to query data in plain English. The initiative addresses years of accumulated data sprawl across dozens of production databases, ClickHouse clusters, Kafka streams, and cloud storage systems, where accessing insights required specialized knowledge of multiple query languages and data sources. Town Lake provides a single SQL interface to all Cloudflare data, with automatic PII detection and role-based access controls, while Skipper layers an AI agent on top to translate natural language questions into accurate, auditable queries that return results in seconds.
The problem was acute: Cloudflare processes over 1 billion events per second across 330+ cities in 120+ countries, but analysts faced tribal knowledge barriers and fragmented credentials to access data scattered across Postgres, BigQuery, R2, and Kafka. Some critical systems relied on external vendors for internal reporting, and sampled data pipelines worked for dashboards but not for billing or security investigations. The solution required building both a unified data layer and an intelligent query interface that could handle fresh, unsampled data for security-critical queries while still serving fast, aggregated data for exploration. Both tools are built on Cloudflare's own infrastructure, including R2 for storage and Workers for compute.
- Both tools are built entirely on Cloudflare's own infrastructure (R2, Workers), removing external dependencies
Editorial Opinion
This is a pragmatic response to a problem that plagues hypergrowth companies: data fragmentation and tribal knowledge. By combining a unified data layer with AI-assisted query generation, Cloudflare is addressing a real productivity bottleneck. The emphasis on security, auditability, and time-bounded access suggests they're treating data governance seriously—a model other large-scale infrastructure companies would be wise to follow. The fact that it's built entirely on their own platform demonstrates both their infrastructure maturity and their ability to eat their own dogfood.
