Over the past six months, the chatbot Claude, from Anthropic, has become my writing assistant with tasks outside of client projects. At this point, however, he is nowhere near becoming eligible for promotion to writing partner. I have serious reservations not just about the quality and consistency of his work product but also about his ethics.

Some days, I’ve considered wearing my bite plane while working with Claude because his outputs cause so much teeth-grinding. Yet I persist because I see glimmers of potential, especially when it comes to developing my own writing skills.

So far, I see the most promise in Claude’s ability to act as a writing coach. I don’t trust him to edit my work or to draft anything but the most formulaic kinds of content, such as announcements about our monthly learning events. But he could have a bright future as a coach, provided we operate with certain safeguards in place.

Here are four ways I’ve been training Claude to help me see my writing from a different angle and improve drafts. Please take careful note of the recommended guardrails!