Guest:
Ellen Brandenburger – Product leader and coach; former head of product at Chegg Skills and Stack Overflow’s data licensing team.
What we cover in this episode:
How Ellen joined Stack Overflow just two weeks before ChatGPT launched, reshaping the company’s future overnightThe creation of Overflow AI: a team tasked with exploring “what’s just now possible” for developersFour iterations of conversational search:V1: a chat UI on top of keyword searchV2: semantic search to handle natural questionsV3: fallback to GPT-4 for gaps in Stack Overflow’s corpusV4: adding RAG for attribution and transparencyWhy attribution and transparency were critical for developer trustHow the team used simple spreadsheets and subject-matter experts to evaluate answer accuracy, relevance, and completenessWhy Stack decided to sunset conversational search despite heavy investment—what they learned and why it wasn’t wastedThe pivot to data licensing: how Stack Overflow leveraged its 14M+ Q&A corpus to power LLM training and benchmarksBuilding industry benchmarks with subject-matter experts to prove Stack data improved LLM accuracy and relevanceKey lessons:
Take one bite of the apple at a time—prototype, learn, iterateProduct in the AI era means managing probabilities, not certaintiesLinks & References:
Ellen Brandenburger on LinkedInThe Changing State of the Internet and Related Business ModelsProLLM: LLM benchmarks for real-world use-cases