{"type":"video","version":"1.0","html":"<iframe src=\"https://www.loom.com/embed/302aab715363403ca86743553ef25768\" frameborder=\"0\" width=\"3342\" height=\"2506\" webkitallowfullscreen mozallowfullscreen allowfullscreen></iframe>","height":2506,"width":3342,"provider_name":"Loom","provider_url":"https://www.loom.com","thumbnail_height":2506,"thumbnail_width":3342,"thumbnail_url":"https://cdn.loom.com/sessions/thumbnails/302aab715363403ca86743553ef25768-5087e363efdc5dcd.gif","duration":1156.585,"title":"How to Prompt Agents With Context","description":"This Loom explains why prompting quality and business context are crucial for getting an agent to perform correctly. The speaker distinguishes context as prompts plus documents, skills, tools, and ecosystem, then demonstrates personalizing a “Hermes” agent by generating a business reference document from Claude or ChatGPT and uploading it under the agent’s Documents tab. They stress that the agent must be told where those reference documents are so it can retrieve them in future chats, showing that it fails in a fresh session without that instruction but succeeds after prompting it to remember the document source. The Loom also covers best practices like using clear, deterministic language and “interview me relentlessly” to trigger clarifying questions."}