This is a simple power move you can implement immediately to save time, preserve mental energy, and serve your future self with excellence.

You return to a project you put on hold six months ago.

The finished document is still there. But the thinking behind it is not.

What feedback changed the direction? Why did you make that final decision?

You may eventually reconstruct the answers, but first you have to do the work before the work.

As I explained in In Pursuit of Elite: Give AI the Middle, Not Your Mind, I use AI as both a thought partner and a doing partner. There are now few meaningful tasks I work on that do not involve AI.

That makes preserving the AI thread an increasingly important part of preserving the work.

In 2012, McKinsey Global Institute estimated that managers and other knowledge workers spent nearly 20% of their workweek looking for internal information or tracking down colleagues who could help with specific tasks.

Saving an AI conversation will not eliminate that entire problem. But it can eliminate one increasingly common version of it: searching for or reconstructing the context behind AI-assisted work.

Most people use AI, get the output they need, and move on. The final deliverable gets saved, but the conversation that helped produce it disappears into a long chat history.

That conversation may contain your source material, instructions, corrections, decisions, feedback, iterations, and final judgment.

The thread is part of the work. Save it with the work.

I generally recommend using a separate AI thread for each meaningful task, decision, or deliverable.

A project can hold shared knowledge, files, instructions, and related conversations. A thread should remain focused on a particular outcome. When the desired outcome meaningfully changes, start another thread.

Anthropic’s guidance notes that irrelevant content can degrade a model’s focus. A long conversation is not inherently bad. An unfocused one is.

Keeping unrelated subjects and angles out gives the AI cleaner context and leaves you a more useful working history.

On the Elite Ops side, I teach a simple rule: If something will take more than 10 minutes, it should generally become a trackable task.

During a Communication Sprint, which may include email, Teams, Slack, or other channels, I often handle an item immediately if it requires fewer than 10 minutes. If it requires deeper work or later follow-up, I move it into my task-management system.

Tracking also gives me a clear picture of work in progress and work velocity: what is active, what is moving, and where work may be slowing down.

When AI materially supports the work, I save the primary thread with the task.

Creating the task and pasting the URL require a little effort. But that discipline can prevent substantial friction when the work returns two weeks or six months later.

The task tells you what must be done. The thread restores much of the context needed to continue.

This is about serving your future self well. When the task returns two weeks or six months later, your future self will be delighted to find the task and thread together, allowing you to rapidly pick up almost exactly where you left off.

The screenshot below shows the Trello card I am using to create this article. The AI Thread field links to this conversation, and the checked “AI thread saved or N/A” item makes preservation part of my definition of done.

The Trello card used to develop this article, with the AI thread and checklist item highlighted.

The system does not require Trello or custom fields. In nearly any task-management system, simply add this to the task description:

AI Thread: [paste URL]

For work I expect to revisit, I generally use the web version of my AI tool because I can copy the conversation URL directly from the browser. I use the desktop version when I need a capability tied to my computer.

One guardrail: Save the private conversation URL for your own workflow. Do not create an accessible shared link unless sharing is intentional, and follow your organization’s policies for sensitive information.

A completed deliverable preserves the result. A saved thread preserves much of how you got there.

Take one minute today and add an AI Thread field or an “AI thread saved or N/A” item to your task template.

One meaningful task, one primary thread. Save the thread with the task.

Special Ops is a month-to-month implementation group for leaders and operators who want to build top-tier learning and productivity systems, strengthen practical execution, and more effectively leverage AI across their workflows and teams. Reach out if you want to learn more.

Serve your future self,

Steve Kahle |  ELITE OPERATORS
In Pursuit of Elite