Create a writing co-pilot

To create a writing co-pilot, you’ll need to upload notes and articles into Claude or GPT before the prompt. This example uses Claude, and is based on Guy Tasaka’s process shared at an “Ask me anything” about AI session.

Inputting data first allows AI to apply the Large Language Model technology to real data.  This could be the reporter’s earlier stories, notes, early drafts, previous articles.

The process is known as R.A.G., or Retrieval Augmented Generation, as displayed in this diagram.

To upload data into Claude, go to Claude.ai and click on “projects” in the middle of the page. You’ll get this screen view:

Beginning uploading data and articles into right section of the project Tasaska inputted ten articles before he enters the prompt for Claude to write the column.

Tasaka noted, “I uploaded 10 articles… Anything I write within that Projects folder is going to look at the writing my streams above, as well as what’s in my knowledge base.”

The full process looks like this: 

1. Create a “knowledge base” in Claude’s project area, with his previous articles, notes, and draft content.
2. Write a system prompt with instructions on voice, tone, and output format, and words to exclude to avoid sounding generated. 
3. Include a specific requests (e.g., “write an outline of an 800-word article”).
4. The AI generates the column based on his knowledge base and prompts.
5. He reviews, asks for clarifications, and requests edits as needed.

The result from the prompt is not always perfect. It tends to exclude brand names, so he needs to clarify the prompt to keep the brand names in the article. Sometimes it will spell-correct brand names, which doesn’t help. But on the whole the  co-pilot delivers teh tone, voice, informed by  background knowledge  and delivers a column that requires less human intervention than starting from scratch.