WorkThe story, in brief
Presentation: AI Agents to Make Sense of Data at OpenAI
OpenAI's internal data agent handles 600+ petabytes. Here's how they solved the context window problem.
The KeyNews take
Why it matters
OpenAI is deploying sophisticated agent architecture internally to manage massive-scale data analysis. The technical approach—MCP, RAG, scoped semantic memory, AST-based grading—signals how production AI systems are being hardened against real-world complexity at scale.
The key facts
6 to knowInternal agent: Kepler (OpenAI's data analyst)
Scale: 600+ petabytes of data
Context window solution: MCP + automated code crawling + RAG
Evaluation robustness: AST-based LLM grading pipeline
Memory approach: Scoped semantic memory for self-learning
Source: Internal presentation by Bonnie Xu at InfoQ (public sharing of internal work)
Go to the source
InfoQ AI/MLinfoq.com
Publisher excerpt: OpenAI’s Bonnie Xu discusses Kepler, an internal AI data analyst agent built to query 600+ petabytes of data. She explains how they overcome context window limits using MCP, automated code crawling, and RAG. Xu also shares how their team leverages scoped semantic memory for self-learning and…