GrowholdThe First AI Tools Ordinary Buyers Ask For by Name
Every AI capability so far has reached businesses through builders. One category is walking in the front door instead, named by the buyers themselves, and the way it arrived tells you how the next one will.
Every AI capability so far has entered the market the same way: through builders. Agents, model integrations, retrieval, all of it gets commissioned by the people making software and reaches ordinary businesses inside someone else's product. We measured that pattern when we found the people buying agents are mostly other builders.
One category is not following the script.
When AI video generation tools get named in demand, the people naming them are overwhelmingly not builders. For every builder asking for one of these tools by name, roughly thirteen ordinary buyers do: media companies, marketing teams, shops, schools. They are not waiting for the capability to arrive inside a product. They already know the tool they want.
Builder-led categories like agents invert the first figure completely.
A whole class rising at once
The second tell is just as unusual. This is not one vendor having a good quarter. Veo, Kling, Seedance, Runway and Higgsfield are all climbing in the same stretch, and the category's combined share of named demand has more than doubled.
Combined share of the class across four consecutive periods, indexed to the first. Individual weeks are noisy; the direction is not.
When one tool rises, that is marketing. When five rivals rise together, a capability crossed the threshold where normal people can use it.
That distinction is the practical one. A single trending vendor tells you about a sales team. A class moving together tells you the underlying capability went from demo to dependable, because buyers who do not read release notes started getting usable results.
The pipeline is rising with it, and voice is sinking beside it
Look at the whole neighbourhood in one picture.
Above 1.0 is gaining share, below is losing it. The smaller names carry fewer signals, so read the pattern, not any single bar.
Three things sit in that chart.
- Premiere Pro is growing as fast as the generators. CapCut too. Generated video is arriving as raw material that still gets cut, graded and captioned, so the editing stack is being pulled along rather than replaced. If you build for video, the money is in fitting the pipeline, not skipping it.
- Voice is moving the opposite way as a class. Every major voice tool is losing share at once, the mirror image of video's arrival. The capability works, which is exactly why it stopped being a destination and became a component.
- Avatar video sits flat between the two, past its novelty, not yet infrastructure.
What to take from it
The buyer-first arrival changes the playbook for anyone building on this capability. When adoption comes through builders, the winning move is APIs and being embedded. When buyers name tools themselves, distribution already belongs to those brand names, and the winning moves are integrating with them, building the pipeline around them, or serving the businesses now producing far more video than their processes can handle.
And the class-rise itself is the reusable lesson. It is the most reliable timing signal we can see: not the loudest launch, but the moment an entire category starts climbing together. That is the market saying the capability is ready, in the only language it speaks, which is what we keep pointing at when we say the market moves at the speed of AI.
