The AI Writer Assistant — why Google Docs was the right surface

Editor’s note, 2026: The prototype has its own case file now — see /work/ai-writer-assistant. This post captures the thinking from when I was first building it.

Most AI writing tools assume you want to leave your current document, go to a new tab, get a block of text, and paste it back. In a drafting-heavy L&D role, that’s a lot of friction for something you do dozens of times a day.

I wanted the assistant to live inside the document. So I built one.

What it does

The AI Writer Assistant is a Google Docs integration that:

• Keeps context. It reads the document you’re in and maintains state across turns — no re-pasting your outline every time you want a variant.
• Works collaboratively. Google Docs is already where drafts go to be argued over. The assistant’s outputs land in that same surface, so the comment-and-suggest workflow everyone already uses just keeps working.
• Speaks L&D. Out of the box it’s tuned for corporate training and instructional design tasks — learning objectives, module outlines, activity drafts — rather than the “write a blog post about SEO” default.

Why Google Docs, and not a dedicated app

Three reasons the dedicated-app instinct was wrong for this job:

1. Distribution. Everyone I work with already has Docs. Nothing to install.
2. Collaboration is native. I’d have to rebuild real-time co-editing, commenting, version history. Docs does all of that for free.
3. Workflow continuity. The drafts don’t leave the system they were going to end up in anyway. That sounds small. It isn’t.

The bigger point

I keep coming back to this: AI tools that win aren’t the ones with the most features. They’re the ones that show up where the work already happens. Context is everything — the model’s context, and the user’s.

More on the full build — the prompts, the Docs add-on structure, what broke and what I rebuilt — in the case file.