From ebooks to template, in one afternoon
Annika has published ebooks on ADHD-friendly nutrition and is building a program to help people with ADHD work on their eating habits. She'd been meaning to turn that material into something people could actually use day to day — a plan, a page, a tool — and it had been sitting on her list for a while.
What moved it forward was connecting Claude to her xTiles workspace on a prior call. Afterward, she uploaded the PDFs of her own ebooks and had Claude pull the content straight into a new project. "I'll just upload my PDFs and you put that in," is how she described the plan — and it worked well enough that by the next time we talked, the import was essentially done. Her own words: "That was so nice, that I did that."
What was left wasn't writing — it was formatting. The knowledge from months of published work was already sitting inside the project; it just needed to look like a template instead of a document dump.
Why it's called an anti-planner
The standard meal-planning advice is a grid: this food on Monday, that food on Tuesday. For a lot of people with ADHD, that grid quietly fails — not from lack of trying, but because a rigid plan adds a decision that has to be defended every single day (do I actually want this today?) on top of the decision of what to eat at all.
Annika's template flips the structure. Instead of assigning specific meals to specific days, it gives you a wide set of good options and lets you pick whichever one matches how you feel that day. No grid to fall behind on, no meal you're now "supposed to" eat because it was Tuesday's square.
Asked directly what makes this different from meal planning in general, Annika's answer was simple: the goal is that you actually eat your food. Executive-function stress around food takes up real brain space, and when that space is freed up, eating gets a lot easier — regardless of where someone falls on the ADHD spectrum.
What's inside the template
The project is organized as a small set of purpose-built pages rather than one long scroll:
- Weekly food plan — the anti-planner itself: big categories of ideas instead of a fixed daily menu.
- Grocery list — built from the same categories, so shopping matches what's actually on offer that week.
- Example plan — a filled-in sample for anyone who wants to follow something ready-made rather than build their own.
- Simple meal ideas — a running list of low-effort options, organized so it's fast to scan when you don't have the energy to think of something.
- Food-likes checklist — a place to mark off what you actually enjoy, so future planning starts from preference instead of a blank page.
The recommendations underneath it are similarly low-pressure: pick a handful of meals to actually cook, and aim for a few that take zero effort. It's a system sized for the days when cooking is not going to happen no matter how the plan reads.
Letting AI do the deciding, when you want that
Because the whole project already lives in xTiles, it can double as context for an AI assistant — not just a static page. Two ways that plays out:
- Auto-build a week. Instead of scrolling the simple-meal-ideas list yourself, ask Claude to pull a handful of options at random and write them straight into next week's planner page as a finished menu. For anyone who finds browsing options draining — or just wants a shortcut some weeks — the choosing happens without you.
- Tailor the template to your taste. Go through the food-likes checklist, select the fruits, vegetables, and proteins you actually enjoy, copy that list, and hand it to Claude with a request to edit your copy of the template — dropping recipes that don't fit and keeping the ones that do.
Neither workflow requires touching a prompt library or learning new syntax. The project itself is the input; you just say what you want done with it.
What gets easier from here
Two things in progress will make this workflow smoother still. A planner-inside-project feature will let a page like the weekly plan reset itself automatically each week, instead of duplicating the page by hand every time. And an MCP update will let Claude edit existing tiles — rearranging or updating a page in place — rather than only creating new ones, plus a smarter default layout so an AI-built page starts out closer to camera-ready.
For now, the pattern already works: bring your own content, let Claude do the first pass, then spend a fraction of the time you'd have spent writing it from scratch making it easy to navigate.