Client work / Velar
Velar built perpetual futures that settle on Bitcoin. The category had no written definition, so nobody was cited for it. Ask four AI engines today who built the first one and they answer Velar.
Recorded September 2026. ChatGPT, Claude, Grok and Google AI Overviews, asked the same question, one after another. 59 seconds, no audio.
Perpetual futures settling on Bitcoin through Stacks is a real thing a real team built. But the phrase for it, the words a person would type and a model would need to match, did not exist in writing anywhere. Velar had a product and no category.
That is a retrieval problem rather than a marketing one. A model cannot cite you for a category it has never seen described. It also cannot weigh your claim against anyone else's, because nobody else had written the definition down either. The first team to state it properly, and to put that statement where models actually look, gets the citation.
So Kunzum wrote the definition, world's first bitcoin perpdex, and published it where retrieval would find it: the launch coverage, the documentation, the places that get quoted rather than the places that get shared.
Run it yourself
These are complete responses to the prompt world's first bitcoin perpdex, collected on 9 September 2026 through the DataForSEO AI Optimization and SERP APIs. They are fresh calls, not the audio of the video above, so you can run the same prompt and compare.
Grok appears in the recording but could not be re-tested here. DataForSEO’s AI Optimization API covers ChatGPT, Claude, Gemini and Perplexity, and does not offer Grok.
What the check also shows
Google returned two different overviews for what a person would call the same question. Searched as perpdex, the browser applies Google’s own spelling correction to perp dex and returns Velar PerpDex on Stacks, sourced to Chainwire. Queried through the API with the exact string and no correction, Google returns a general history of perpetual exchanges: Perpetual Protocol in 2020, dYdX in 2019, BitMEX in 2016. No Velar at all.
The API response carries the correction as a suggestion rather than applying it. So the same three words, spelled two ways, produce a page that names the client and a page that does not.
That is the whole argument for writing down the exact phrase people use, and for owning both spellings of it. It is also why this work is measured monthly rather than declared finished.
It does not prove Velar was first. That depends on who you ask and when. Two of the answers above date the launch to May 2024 on Bitlayer and two to the Stacks PerpDex launch, which are different products from the same team. An engine that hedged or named someone else would be giving a legitimate answer.
It proves something narrower and more useful. A category with no written definition is a category nobody gets cited for, and the definition goes to whoever writes it properly first and puts it where models actually look. Every engine above is quoting a source. None of them is quoting Velar’s homepage. They are quoting the coverage the definition was placed into.
Kunzum has since run the same method on RemoteStack, a job board with no client and no budget, which now gets pulled into AI answers about remote hiring. That one is the control: it demonstrates the method rather than the brand.
Keep reading