Recall AI just launched a beautiful redesign of their notebook editor, complete with slash commands, tables, and—finally—clickable checkboxes that feel as satisfying as Notion’s. The interface is genuinely impressive.
But there’s a critical problem hiding beneath the polish: the AI can’t recognize which boxes you’ve checked.
Where This Breaks Down
I discovered this while testing my journal workflow. Previously, I used markdown formatting (- [ ] for unchecked, - [x] for completed tasks), and asking the AI “what tasks haven’t I finished?” worked flawlessly—it understood the syntax and gave me accurate results.
With the new visual checkboxes? Everything changes. When I query the same journal, the AI lists every single task, regardless of whether I’ve checked it off. It’s completely blind to your checkbox status, even though Recall exports the proper markdown behind the scenes.
The Core Issue
There’s a disconnect between what you see (a nice visual interface) and what the AI actually processes. The visual layer is divorced from the AI’s parsing capabilities, leaving your task queries useless.
The Practical Takeaway
If you rely on AI to help you manage your workflow—whether that’s tracking what you’ve completed, finding open items, or running queries against your journal—stick with the markdown format for now. The new checkboxes are polished and attractive, but they’ll sabotage any AI-dependent workflow until Recall solves this.
What’s Next
This feels like either an oversight that’s fixable or a limitation that needs transparency. Either way, the Recall team should address it soon. In the meantime, I’d love to hear if you’ve found a workaround—drop a comment below.
See the full side-by-side comparison in the video
