We built a library of slide layouts at work — 154 of them, 41 themes, so ~6,150 combinations. Every single file is served publicly. No auth. 193 HTTP requests and you could mirror the whole thing.
I found that out when someone pointed it out to me, and my first instinct was to lock it down.
I didn't, and here's the reasoning I'd like a second opinion on.
What a mirror gets you is parts. What nobody can copy by downloading files is the assembly — searching the catalogue by intent, working out the order of twelve slides so they don't repeat, keeping one theme across the whole deck. The inventory was never the moat. It's the discovery surface, and locking it kills the only free acquisition channel I have.
So the paid line moved. Free: take as many individual layouts as you want, forever. Paid ($9.99/mo): the model composes the deck for you over MCP.
Two things I got wrong on the way here:
I thought "nobody measures whether the text actually fits the slide." Four products do, and one of them sits in the same MCP channel I was aiming at, for free.
I also thought a competitor didn't sell to individuals, and built a whole positioning argument on it. The source was a 2019 Reddit comment that had already been corrected in its own thread. They sell single seats at $28.60/mo.
Revenue is zero so far — the paid tier just went live. Posting this at the "before it works" stage on purpose.
(Worth saying: this is a company product, not a solo side project. The decisions below are mine to argue for, but the runway isn't my own savings.)
The question I keep circling: when your inventory is copyable but your composition isn't, is the free tier acquisition or just leakage? Anyone here drawn that line and regretted it?