I recorded a client meeting, handed the AI-generated summary to an agent, and asked it to build a project plan. What came back was organized and mostly right, but it was missing context that would have made it better.
What Happened
The recording tool did a solid job generating a summary with action items and key decisions. On paper, the notes looked complete.
But when I passed those notes to an agent, the output had gaps. Priorities the client emphasized were condensed. Concerns they raised in passing weren’t captured. Context around why certain decisions were made wasn’t part of the summary. The agent only knew what the summary told it.
Why Summaries Aren’t Enough on Their Own
AI meeting summaries are designed to compress. They take a 45-minute conversation and reduce it to bullet points and action items. That compression is useful, but it means some things get left on the table.
Offhand comments that turn out to matter don’t always make the cut. Nuance gets condensed into clean statements. When another AI agent picks up those notes downstream, it treats every line as equally weighted. It doesn’t have a way to tell the difference between a firm commitment and a passing thought.
The Fix
Keep the full transcript. Every time.
The summary works well as a quick reference for yourself. But any time you’re passing meeting context to an AI agent, give it the transcript alongside the summary. Let the agent work with the full conversation instead of a compressed version of it.
I now store both the summary and the raw transcript together. When an agent needs meeting context, it gets both. The output is better for it.
The Broader Principle
This applies beyond meeting notes. Any time you’re feeding AI-compressed content back into another AI system, you’re compounding information loss. Each pass leaves a little more behind.
Keep your source material. Storage is cheap. Losing context in a client relationship is not.