Skip to content
Help Center
Try Draftbit Free

Using tokens efficiently

Optimize your daily token usage to reduce waste

Knowing how to use tokens efficiently is key to getting the most out of Draftbit. Read this guide to learn how to use tokens efficiently and keep your costs down.

Do the easy stuff yourself

Section titled “Do the easy stuff yourself”

The first and most obvious way to reduce your token usage is to do the work yourself. If you need to make a small change like adjust the padding of an element, you can easily do that yourself without needed to spend tokens having the AI do it. Use the Chat mode to ask the AI for guidance on how to do it yourself which uses very little tokens compared to having the AI code for you.

Stick with small, focused changes

Section titled “Stick with small, focused changes”

The bigger the request, the more work the AI will do without interruption. Keep your requests focused on a single task that you can incrementally refine through successive requests. This will help you keep better track and control over how fast your tokens are being consumed. Read more about composing task prompts.

Avoid repeated automated error fix attempts

Section titled “Avoid repeated automated error fix attempts ”

Asking the AI to fix an error multiple times in a row can waste tokens unnecessarily. After each attempt, pause to assess the results and refine your next request accordingly. Some programming issues may be beyond the AI’s ability to fix automatically. In those cases, it’s often more effective to research the issue yourself and make manual adjustments before trying again. If you’re unable to get unstuck, reach out to our support chat in the bottom-right of the Builder.

Add error handling to your project

Section titled “Add error handling to your project ”

If your project keeps running into the same error, try asking the AI to strengthen error handling and insert detailed logging in the affected code. The AI is especially good at adding comprehensive error logs—even between specific functions or key steps. These logs not only help pinpoint what’s going wrong, but also provide the AI with clearer diagnostic information for future fixes. With better visibility into the issue, the AI can make more targeted and effective corrections.

Rollback changes that aren’t working

Section titled “Rollback changes that aren’t working ”

You can restore your project to an earlier state without using any tokens—functioning like an “undo” button. This can be a real time-saver when things go off track. However, keep in mind that there’s no “redo” option: once you roll back, all progress made after that point is permanently lost. Make sure you’re ready to commit before you use it.