Everyone who works with AI is now a manager
How to brief AI and check its results without rewriting everything
We have published a blog post by Aleksandra Albera on why checking AI results often takes as long as writing the document from scratch, and how to change that
Four principles instead of rewriting
The post organises work with AI the way a manager organises a team’s work: a brief with context and acceptance criteria, limits set before the work, acceptance based on four fixed questions and templates captured in the procedure, not in the chat history
Three prompts to copy
The text contains three ready-to-use prompts: demanding a source for every claim, a critical review by a second model and a source audit in a table that shows whether the text says more than the source. We also describe how we built this way of organising work into SAIE: into PROXY:AI and TWIN:DESK
The full post with prompts to copy
