Reader situation
AI summaries make it easy to catch up on a discussion. It’s also easy to respond by repeating what the summary said. A contribution that repeats the summary adds little. One that checks the context and adds your own experience moves the discussion forward.
Central idea
A discussion summary is most useful when it helps a member return to the conversation with an informed contribution. Verify the context, identify any omission and add your own experience rather than repeating the summary.
Worked example
Fictional example
The people, discussion and summary are invented to illustrate an approach. The summary is illustrative and doesn’t describe how any specific product generates summaries.
Labeled fictional discussion
In a Circle for operations leaders, members discuss whether to bring outsourced customer service back in-house. Over a week, twelve members post. Several describe moving work in-house and improving quality. One says costs rose sharply. Another explains that they returned to outsourcing after two years. A third asks whether anyone has tried a hybrid model, and no one answers. The thread is long, and the replies vary in detail.
Illustrative summary
Felipe returns after a week away and reads a summary:
Members discussed bringing customer service in-house. Most reported improved quality. Some noted higher costs.
Context or omission check
Felipe reads the original thread before replying. He notices three things the summary left out:
- One member reversed the decision after two years, which suggests the benefits may not last in every situation.
- The cost increase was described as significant, not minor.
- The question about hybrid models went unanswered.
The summary wasn’t wrong, but it made the discussion seem more settled than it was. If Felipe had replied based on the summary alone, he might have added another voice in favor of moving work in-house, without engaging with the members whose experience was different. He also checks who said what, so that he can address the right person.
Personal contribution
Felipe has experience with a hybrid model. He replies to the unanswered question:
Responding to the question about hybrid models: we kept complex cases in-house and outsourced routine requests. Quality on complex cases improved, and costs stayed manageable. The hardest part was handing cases between teams, which took several months to get right. I’d be interested to hear from the member who moved back to outsourcing. Was cost the main reason?
His reply adds new experience, addresses a gap and engages directly with another member’s view. It also includes a difficulty, the handover between teams, which gives others a realistic picture rather than only the benefits.
Avoid unsupported feature claims
In his reply, Felipe doesn’t mention the summary or describe what it did. He responds to the people in the discussion. When writing about AI tools, it’s best to describe your own experience rather than making claims about how a tool works or what it can do. Claims about a tool’s behavior can mislead other members, especially if the tool changes or works differently for them.
Boundaries
- Summaries can contain errors. Check the original before relying on a point.
- Don’t quote a summary as though it were a member’s words.
- Don’t share summaries of private discussions outside the space where they took place.
Reflection questions
- When did you last respond to a discussion based on a summary alone?
- What did the original contain that the summary didn’t?
- What experience of your own could you add to a discussion you’re following now?
One practical application
Choose one discussion you’ve caught up on through a summary. Read the original, find one gap or open question and respond with your own experience.
