We started Plain in April 2026. The trigger was very concrete — for half a year we'd kept running into the same small thing: AI could already draft a respectable deck outline, but actually turning it into a .pptx and sending it to the client took five copy-pastes, three font-size tweaks, an alignment pass, a stock-photo hunt, and a last worry about whether the client's Mac would even open it. None of that is about AI, yet it ate 90% of the time.

We didn't want "one more AI tool." We wanted to shrink that gap between "thought it through" and "shipped it" until a single link replaces the .pptx attachment.

A document shouldn't be a file. It should be a link — clients, teammates, AI all opening it and seeing the current version.

So Plain's source is Markdown / CSV (diff to git, edit through AI), and what gets rendered is HTML (read by humans, exported for legacy systems). It's the harder route — we have to write our own marp themes, our own visual-edit bidirectional mapping, our own @ref linking — but it gives us the one thing that matters: your content stays editable forever, rollback-able forever, re-interpretable by AI forever.

We promise three things:

1. Transparent pricing — billed by AI tokens, not by seat. One heavy user shouldn't cost as much as a team.
2. Always exportable — .pptx / .docx / .xlsx / .pdf will always be free. We don't make money off export lock-in.
3. We don't hoard data — your content lives where you choose (cloud or local). We're just the tool. Delete your account and it all leaves with you.

The team is still small. If you're aligned with all this — email hi@inplain.app or find @_zephyr_dev on X. Talking beats writing requirements docs.