A Great Research Programme and a Terrible Sales Channel

We've turned off self-serve signups on Hyperframe. Access runs through a demo for now, not an open account page. Before you file that under "startup quietly retreats," it's worth seeing what the free tier actually taught us, because the lesson turned out to be worth more than the traffic was.

How we ended up with thousands of accidental users

A naming coincidence did most of the work. Shortly after our public launch, HeyGen, a well-known AI video company we have nothing to do with, released a product called Hyperframes. With an 's' at the end. We were here first, for what that's worth, which is nothing. It's an MCP-style HTML video builder, nothing like what we make, and the traffic hunting for it landed on us instead.

We finished up with thousands of free accounts making thousands of videos. So now I can tell you roughly what a spelling coincidence is worth in signups. What it's worth in customers is a different number, and we'll get to it.

What people typed into the brief box

You learn a lot about who's shown up from what they ask for. A sample from the actual list: "The Printer Race," which was a race to establish which of two printers is the better printer. A brief asking us to debunk the myth that sheep are herbivores and wolves eat sheep. "Alien Tries Earth: Instant Regret Edition." One person pasted the full lyrics of a children's bath time song into the box and submitted that. "Understanding the Clitoris" which I hope you can figure out by yourself.

I have questions about several of these, starting with the sheep.

None of it is what Hyperframe does. We match a script against a specific, licensed, approved footage library. We don't invent footage from a prompt, so a brief for something that exists in no library was never going to produce anything useful, however good the engine underneath.

There was a quieter version of the same confusion, and it taught us more. People kept typing scripts as stage directions. "The sun rises over a mountain," entered word for word, expecting the system to render that exact shot. What they wanted was a tool that turns a sentence into footage from scratch. That category has a leader, and in March OpenAI announced it was discontinuing Sora, the app most of them were describing. Reporting at the time put its running cost near a million dollars a day against roughly $2.1m in lifetime revenue.

We don't have a million dollars a day. If we did, I can promise we wouldn't spend it persuading people the earth is flat (a genuine prompt someone typed into Hyperframe). The best-funded version of the thing they were asking for couldn't make that work, but no onboarding flow of ours was going to talk them out of wanting it either.

The free tier was genuine research

It gave us real value, to be fair. Thousands of videos stress-tested our infrastructure and surfaced bugs at a scale our own testing never would. People tried the product in other languages, unprompted, which is now a live signal for something we're weighing up.

Watching people arrive with everything from a finished script to a half-formed outline to a bare topic pushed us to support that whole range properly, instead of assuming everyone turns up with a script in hand. That flexibility came straight out of watching free users struggle without it.

So the free tier was a decent research programme. It was just a terrible sales channel.

Enterprise buyers don't sign up. They sneak in on Gmail.

What actually closed the door: the people we build for, teams inside large organisations, showed up too. They just didn't use their work email.

Every so often a generic email address turned up attached to a name we recognised from a meeting, someone we'd sat across a table from a week earlier. You notice that. We watched people from big companies come in on personal Gmail addresses, because running unapproved software on a corporate account is a governance question, and governance is exactly where enterprise adoption decisions get made or stalled. The person who wants the tool can't adopt it alone. The person who could approve it doesn't browse signup pages. Selling to an enterprise is a different motion from collecting individuals, and an open form serves the second while quietly lying to you about the first.

Open trials make sense for a broad tool where nearly anyone who signs up is a plausible buyer. We're the opposite: a deliberately narrow product for enterprise teams, narrower still since explainer video became the flagship. Most of the volume a free tier pulls in was never going to buy, and the signal it throws off is mostly noise wearing a traction costume.

What replaces open signup

Hands-on access isn't going anywhere. Demos and pilots still put the real product in a prospective client's hands, on their own footage, with their own governance questions answered up front. That's how these organisations actually buy. What's gone is the ability to arrive unannounced and start generating video with no qualification at all. The open signup page is off for now, not for good. We plan to come back to it with an actual plan.

The bath time song was genuinely catchy, though. We checked.

If you fancy trying Hyperframe, book a demo with me. Or just email hello@hyperframe.ai.