Give the new idea its own path

Yesterday, I wrote about AI answers that should disappear.

Sometimes though, ongoing work produces a new problem, risk or deliverable worth pursuing. That work needs its own conversation, but it still needs the context that produced it.

That is what Fork is for.

Fork creates a new conversation from an AI turn. Everything before that point comes with it. From there, both conversations continue independently.

There are plenty of practical uses:

• While preparing a launch checklist, supplier delays emerge as a risk. Fork the conversation to build a contingency workflow without derailing the launch coordination.

• Weekly planning exposes a recurring bottleneck. Fork it to build an SOP while the original conversation finishes the priorities, owners and deadlines.

• While building a feature, a potential bug appears. Fork the conversation to investigate it without interrupting the current implementation.

• Closer to home: once an article’s direction is settled, fork the discussion to develop the thumbnail while the original conversation finishes the call to action and publication workflow.

As with Side, Claude Code uses a different name for the same pattern: /branch.

The practical rule is simple:

• If the thought only needs an answer, use Side.

• If it deserves its own outcome, use Fork.

Same starting context. Different destination.

If you want help making better use of the AI tools you already have, without adding more complexity: let’s talk.