The limiting factor—how to design an AI software factory for speed | Geoff Charles (Ramp CPO)
Ramp Chief Product Officer Geoff Charles outlines how product organizations must transition from focusing solely on individual output to engineering an AI-powered software factory. Drawing parallels to Formula 1 racing, he explains how automating one part of the product lifecycle inevitably exposes new bottlenecks in discovery, definition, review, testing, and coordination. Charles details the internal AI agents Ramp built—including Glass, Inspect, Review Buddy, Testo, and Gadget—and outlines the three evolving archetypes for product managers in an AI-native world.
- Ramp built an internal AI agent called Glass that connects directly to Snowflake data, customer research repositories, design systems, and the production codebase. Read →
- Dumping raw feedback into large language models fails at scale: a one-million-token context window holds less than 0.5% of Ramp's Gong call transcripts. Read →
- Ramp treats internal employee inquiries as programmatic API calls rather than manual interruptions for product managers. Read →
- In Formula 1 racing, the driver accounts for only 15% of the impact on winning; the real gains come from optimizing the interaction between the car, pit crew, and team. Read →
- Ramp uses an internal tool named Inspect with over one million sessions, allowing non-engineers to submit 1,000 pull requests per month. Read →
- Ramp built dedicated internal AI agents, including Glass, Inspect, Review Buddy, Testo, and Gadget, to handle routine tasks across triage, testing, and review. Read →