Why we're in a hurry

Anything this different gets copied. That was never up for debate.

What I can do is tell you what I actually know, because I'd rather you hear it from me than work it out later and wonder what else I skipped.

We ran a scan. July 12 to September 14 this year, 64 days, about 30 searches plus a direct read of product pages, GitHub, the Product Hunt leaderboards and the vendors' own release notes. If you want the source list, write to support and it's yours, it's a real list. And treat the count as a floor. Nobody reads every feed.

What surfaced in 64 days

Five new products or features that touch what we do. Five in nine weeks. Roughly fifteen credible products in total now touch some part of the job.

How many do the whole job? Meaning: learn your voice by talking with you, hand you a document you own and can take anywhere, run on the AI you already have, carry twelve years of judgment about what actually gets read, and charge once.

Zero. I checked twice, because I wanted the answer to be zero and that's exactly when you should check twice.

The market has split into two halves that haven't met. One half interviews you and then keeps what it learned inside their subscription. The other half hands you a document you own but builds it from writing samples and has no opinion about whether the writing works. Neither half is in any hurry to become the other, because their business model is the reason they're the half they are.

So why the hurry on my end?

One of the five. On September 7 OpenAI shipped a feature called Writing Styles. It reads your connected Gmail, Drive, Slack and SharePoint and learns your phrases and sign-offs on its own. Right now it's on ChatGPT Work plans only, not in the ordinary chat most people use, not on free, and what it learns stays inside ChatGPT by design. That's the one that matters, because it's the only one that could make the problem go away for a lot of people instead of competing on the fix.

Will it? My honest read: it learns the voice you use in your sent folder, which for most people is the voice that says "per my last email." It has no idea what gets a bill reversed or a listing read. And it doesn't let you leave with anything. Those are three real gaps. They're also gaps a very large company could close if it decided to.

Our best guess at the window, and it's a guess with homework behind it, is somewhere between nine and fifteen months. Bounded on one side by that feature reaching the free tier and actually working. Bounded on the other by somebody with a real audience bolting a spoken interview onto a tool that already has the judgment part. There's one company well placed to do that, and conveniently, they just moved their business toward monthly subscriptions, which a one-time product would undercut. That conflict is a better moat than any technology I've got.

What I'm not going to do

Give you a launch date or a promise. I've done that before, too early, and I ate it. Rushing to be ready and rushing to sell are different things, and I got them confused once.

What I am going to do is keep building the part that's hardest to copy. The tech, sure, somebody will get there. Twelve years of knowing what a person actually says yes to, and the taste to leave the rest out? That isn't on GitHub.

And your voice? That one can't be copied by anyone. Not by me, not by them. It's the only part of this that was never for sale.

Does the hurry make sense now? It's not fear. It's a window, and I can count.

Five copies in nine weeks, zero doing the whole job, and a clock I can hear.

If you'd like to know what happens next, the form below is where I'll say it. No dates, until there are dates.