Talking to your computer is becoming ordinary, but not everywhere

Something is shifting. Wispr Flow has built an audience of people who no longer type but talk. Google announced Rambler at its Android event in May: dictation inside Gboard, with Gemini behind it, taking the "ums" out and following a correction you make halfway through a sentence. And OpenAI says it will show its first device later this year - screenless, with Jony Ive involved, and your voice as the way you work it.

That is no longer a trend, it is a direction. We believe in it too, or we would not be here.

The friction is not the technology

Practice, though, says something the demos do not. The hard part of speech is rarely the recognition; it is where you are standing.

Not every public place lends itself to saying something out loud. On a train, in an open-plan office, in a waiting room: nobody is going to dictate an account number, a case reference, or the reason they were off work last week. Not because it would fail, but because the person beside them is listening. A keyboard is silent, and that is a feature.

That is the brake the whole field is running into, and a better model does not take it off.

The narrow strip where it does fit

So the niche we serve is a deliberate one: people who have to leave something on a website.

A question for support. What went wrong with an order. An explanation attached to an application. That is personal, but it is rarely the most sensitive thing anybody has to say, and it is usually the slowest box on the page, the one where people stall halfway and simply never send the form.

That is exactly where speech is a relief rather than a risk. It helps that nothing anyone says is kept: the recording is transcribed and thrown away inside the same request. There is no place for it to end up.

And the desktop has been left behind

There is a second reason we think this is not settled. Phones get better at voice input every year, and the announcements above are largely about phones. On a desktop the story is a different one: dictation there is still uneven, buried in the operating system's settings, and far from equally good in every box.

A form on a website is filled in on both. So it has to work on both, and it has to work without your visitor having to install or switch on anything first.

We believe in speech; the user decides

Which is to say: we think there is room for competition, and we are glad to be playing into a market that is still taking shape.

We believe in speech to text. But what it is worth is decided by the person using it and not by us: whether they actually press that button, and whether what comes out is better than what they would otherwise have typed. That is not a question you answer from behind a desk.

From the first of November

We hope this tool has been built at a good moment for a fair few websites, and we would like you to try it yourself. From the first of November you can, for nothing.

We look forward to having you as our first customer, including if you pick the free Hobby plan.