The Real Problem With AI? It's Us. (And this feature fixes it)

Most people still type keywords into AI like it's Google, even at AI companies. So how do you get good answers when people don't know how to ask good questions? Eudald Camprubí, co-founder of Nuclia and Software Fellow at Progress, explains how agentic retrieval in Progress Agentic RAG uses three models, each with its own job: 1. Rephrase: turns a messy question into a clear one the AI can understand 2. Executor: plans which sources to search, in what order, and what to ask each one 3. Summarize: combines what it finds across all sources into one final answer You can choose a different LLM for each step from 50+ models, so you control the balance of quality, speed and cost. Full session: https://www.progress.com/resources/webinars/why-good-ai-still-gives-incomplete-answers-and-how-to-fix-it