Mem Tutorial: How to Create a Spark File to Capture Ideas in Mem

Capturing ideas consistently is the difference between having inspiration and turning it into output—and Mem’s “spark file” setup is presented as a simple system for building that habit. The core idea comes from writer Stephen Johnson: a spark file is a loosely structured document where hunches and raw material live until they’re ready to become articles, speeches, software features, startups, or even whole books. Its organizing principle is intentionally minimal—entries are kept in chronological order with no heavy taxonomy—so the system stays easy to maintain while still preserving the timeline of creative momentum.

In Mem, the same approach is recreated through tagging rather than a single rigid document. The key move is creating a tag named “spark file.” Once that tag exists, anything tagged with it automatically appears in the spark file view in chronological order. That means the “structure” Johnson describes is effectively handled by Mem’s tag-based organization: users don’t need to decide categories up front, and they don’t need to stop to file ideas perfectly. The workflow is demonstrated in two capture paths.

First, ideas can be captured via text messages sent from a phone to a Mem number. When an idea strikes—whether it’s a thought while listening to a podcast or a concept encountered during daily life—it can be texted to oneself and then tagged as “spark file” so it lands in the chronological collection. A concrete example uses a podcast quote from Sunil Gupta’s book “Backable,” where the insight “transformed his to-do list into a to learn list” becomes a tagged spark entry. Another example imagines an app concept—“distraction free writing app that leads to flow”—and shows how it would appear under the spark file tag.

Second, Mem Spotlight can capture ideas directly from reading. After selecting a nugget from an article (the example references a recent Medium piece by Ryan Holiday), the user invokes Spotlight with Command-Shift-Space, confirms the capture, and tags it as “spark file.” Returning to the spark file tag then reveals a chronological list of ideas gathered from multiple sources, mirroring Johnson’s approach.

The practical takeaway is about filtering: a spark file should have the least filtering possible. It’s framed like the top of a funnel—where many raw leads are collected before any later sorting or refinement. Since ideas rarely arrive in neat, linear categories, the system’s low-friction capture and chronological ordering are positioned as the best way to avoid losing inspiration and to build a usable reservoir for future writing and product thinking. A link to Stephen Johnson’s article is mentioned as additional reading, with an invitation to ask questions if needed.