I came across a post on Hacker News this morning that stopped me mid-coffee. The founder behind bankstatementconverter.com wrote a blog post titled, plainly, "My Business Is Dying." No spin, no silver lining dressed up as a growth hack. Just a founder watching his one useful thing — converting bank statements into usable data — get absorbed into the general-purpose capability of large language models, for free, as a side effect of someone else's product roadmap.
If you've built a single-purpose SaaS tool in the last decade, read that post twice. Then read it again after you've stopped feeling smug that it's not you. Yet.
The Trap of the Single Clever Trick
His business worked because he solved one annoying problem better than anyone else bothered to. That's a perfectly good way to start a company. It's how most of us start — you scratch your own itch, someone else has the same itch, you charge them a fiver a month, and suddenly you've got a business.
The trouble is that "one clever trick, well executed" is exactly the shape of problem general-purpose AI is devouring right now. Extracting structured data from a messy PDF? Summarising a document? Converting formats? Rewriting text in a different tone? These were businesses. Now they're a paragraph in a system prompt.
I've built things in that exact category. RSSMasher, Article2Video, BookMasher — each one takes raw content and turns it into something more useful, more polished, more monetisable. Digital alchemy, as I like to call it. And every single one of them is, in some dimension, a "clever trick" that a sufficiently capable model could theoretically do unassisted if you asked it nicely enough and fed it the right instructions.
That's not a comfortable thing to admit. But pretending otherwise is how you end up writing your own "my business is dying" post in eighteen months' time.
Defensibility Was Never About the Trick
Here's the bit worth sitting with properly. The bank statement converter founder's mistake — if you can even call it a mistake, hindsight is cheap — wasn't picking a niche that was too small. It was that the entire value of the business lived in the trick itself, rather than in everything wrapped around it: the workflow, the trust, the integration, the customer relationship, the accumulated understanding of what this specific type of user actually needs on a Tuesday morning when their accountant is chasing them.
A model that can extract numbers from a PDF is not the same thing as a product that a bookkeeper trusts enough to feed their client's finances through without checking every line. The gap between "can technically do it" and "is the thing people reach for without thinking" is where defensible software still lives. It's just a much narrower gap than it used to be, and it's closing.
This is exactly why I've never let the Masher suite stay as one tool doing one job. RSSMasher isn't valuable because it can turn an RSS feed into content — any model can rough that out now. It's valuable because it plugs into a workflow someone already has, with the scheduling, the formatting, the multi-platform output, the bits that make it a system rather than a party trick. The trick was always the entry point, never the moat.
Diversify the Surface Area, Not Just the Product List
The obvious lesson is "build more than one tool," and that's true as far as it goes — it's why AIMasher and MarketMasher and VidMasher exist alongside each other rather than me betting everything on one horse. But the deeper lesson is about where you build the moat within each product.
Own the workflow, not the function. Own the data and the history a customer accumulates with you, not the algorithm that processes it. Own the trust that comes from being right ninety-nine times out of a hundred in a specific domain, not the cleverness of the underlying transformation.
If your business is one prompt away from being a feature in someone else's chat interface, that's worth knowing now, while you can still do something about it — expand the surface, deepen the integration, get closer to the actual human problem rather than the technical one.
The founder at bankstatementconverter.com is being honest in public about something most of us would rather not think about. That's worth more than another post telling you AI is going to change everything. We know. The question is whether your business is the trick, or the thing built around it.
— Wayne