AI review step

We have a draft blog post, so let’s get an AI model to review it. We’re going to continue the pattern of incremental improvement by asking the AI model to provide feedback on the draft blog post, but stop short of making any changes.

Click the (+) handle to the right of the draft blog post block to create a new AI block and enter the following instruction:

Review the draft above. Suggest a handful (3 to 5) high priority, impactful changes I could make to improve it. Don’t rewrite it, just suggest and explain potential improvements.

Here’s what your canvas should look like at this point:

Finalizing the blog post

We have a draft and a set of suggested changes from the previous steps in this guide. Now, we’ll add a block to implement the suggested changes and finalize the blog post.

Click the (+) handle to the right of the review block to create a new AI block. This will be our ‘refine’ block.

Before adding instructions to our refine block, we need to make sure it has access to the draft blog post. To do this, we’ll connect the output of the draft blog post block to the refine block. Click and drag the output handle of the draft blog post block, hover over the refine block and release to connect. This will create a connection between the two blocks, passing the draft blog post to the refine block.

You can move blocks around to get a better view of the connections. Here’s what your canvas should look like after connecting the draft blog post to the refine block and optionally adjusting block positions for clarity of flow:

Now, enter the following instruction for the refine block:

Write the final version of the blog post, incorporating the suggested changes from the review block. No preamble, just the final blog post. Use markdown formatting.

Congratulations! 🎉

You’ve successfully created a canvas that generates a blog post outline, writes a first draft, reviews the draft, and refines it into a final version, which should look something like this:

Learnings so far

By following this guide, you’ve learned:

  • How to create a canvas in Hunch
  • How to add blocks to a canvas, instruct and execute them
  • How to access reader view to review block outputs
  • How to connect blocks to pass data between them
  • The importance of incremental improvements in AI workflows

All of that in a few minutes! 🚀

Next, we’re going to further enhance our blog post by generating and adding a hero image.