A daily heartbeat should make decisions, not just summarize links

Back to blog

A Daily Heartbeat Should Make Decisions, Not Just Summarize Links

AI makes summarization cheap.

That does not mean every summary is useful.

A daily heartbeat can summarize inboxes, meetings, feeds, project notes, browser tabs, and tool updates. But if the output is only a cleaned-up pile of information, the operating problem is still there.

The better question is: what decision does the heartbeat help make?

The Problem With Passive Digests

Passive digests are interesting, but they often stop at description:

• here are the links
• here are the updates
• here are the notes
• here are the highlights

That can help, but it does not necessarily tell anyone what to do next.

The Better Loop

A useful heartbeat needs three steps.

1. Capture The Signal

The system should record where the signal came from: local work, meeting notes, Airtable records, public articles, user notes, or tool updates.

The source trail matters because not all signals deserve the same confidence.

2. Score The Signal

For a content heartbeat, score each signal for usefulness, freshness, proof, story fit, CTA fit, safety, recordability, and trend fit.

The scores are not magic. They make the decision visible.

3. Route The Next Action

Some signals become videos. Some become blog posts. Some become project tasks. Some get parked. Some should be rejected because they are too risky, too private, or not useful enough.

That is the difference between a digest and a workflow.

Use The Content Heartbeat ICM Filesystem

The Content Heartbeat ICM Workspace Template is a small workspace for running this loop without starting from a blank page.

Open the Google Drive template folder here: Content Heartbeat ICM Workspace Template

It is designed to be approachable in normal working files: a start-here guide, source map, review gate, example signals, ranked ideas, and prompts. The point is not to make the work feel technical. The point is to give an AI assistant a workspace it can interpret without burying every rule in one long prompt.

Use the filesystem this way:

1. Start with the start-here file so the purpose, cadence, and review rules are clear.
2. Keep the source map current so the assistant knows what inputs are approved.
3. Capture one signal at a time with enough source context to inspect later.
4. Score the signal before deciding whether it deserves action.
5. Route the next action into a draft, task, resource, rejection, or parked idea.
6. Review the queue before anything gets published or sent.

The practical test is simple: after one heartbeat, there should be a clearer idea queue, not just cleaner notes.

If ICM is new, the broader explanation is here: What Is ICM?

You can also learn more from the Clief Notes Skool Community, where Jake Van Clief and the community discuss ICM and AI workspace patterns.

The Practical Takeaway

If an AI system produces lots of updates but not clear decisions, it may be summarizing instead of operating.

A real heartbeat should help answer:

• What matters?
• Why does it matter?
• What should happen next?
• What needs review before action?

Start with the workspace, run one heartbeat, and look at the idea queue. If it gives clearer decisions instead of just cleaner notes, it is on the right track.