In this AI Workshops session, Forum3 AI specialist Kait Smith explains the difference between memory, instructions, project context and the current conversation in ChatGPT and Claude, and why each important fact should have one clear home. She shows where to turn memory on, how to ask an AI what it remembers about you, and how temporary chats keep one-off or sensitive work out of memory.
The session also covers connectors in Claude and plugins in ChatGPT, including why read and write actions deserve different levels of scrutiny, a 60-second check before important work, and a live comparison of dictation and voice mode.
Viewers leave able to decide what belongs in memory versus the current chat and to check what information their AI is using.
Memory + Context is a two-page Forum3 guide to keeping an AI informed without filling every prompt with your life story. It distinguishes five places information can live: memory, instructions, project or workspace context, the current conversation, and connectors or plugins.
The guide recommends giving each important fact one clear home and naming a source of truth when versions conflict. It includes a 60-second context check for important work, a prompt that has the AI list its sources and flag anything outdated, contradictory or missing before it begins, and a note on using temporary or private conversations for one-off or sensitive work.
Readers can use it to work out why an AI is relying on the wrong information and fix where that information lives.
Built for restaurants. Proven across consumer brands. No pitch deck, no pressure. Just a conversation about where you are and where AI can take your business.
