Versely
    Guides

    Crypto explainers: finance presenter rules

    No synthetic market hosts. Diagrams, captions, founder footage you have.

    Versely Team••4 min read

    No synthetic market hosts. Diagrams, captions, founder footage you have. That is the whole production rule for a crypto explainer that still wants to earn.

    YouTube's inauthentic-content third bucket is AI personas as human experts on health, legal, finance, or politics. An invented "analyst" walking a viewer through an allocation is the example Help wrote. A talking avatar diagnosing a token is the same shape wearing a ticker. Do not build the channel on that presenter.

    The host is the kill, not the ticker

    Help's heading is "AI Personas Related to Sensitive Topics." Three pieces have to land together: a synthetic persona, a named domain (finance is named), and the framing of a human expert talking to the viewer. Take any one away and you are outside the heading. That is why a mascot on a comedy channel is not this policy, and why a credentialed founder using editing tools on their own video is not this policy either.

    What fails:

    • A generated face styled as an adviser, "host," or desk talent telling the viewer what to buy, stake, or rotate.
    • Implied credentials: office set, name plate, "as a trader" in the first line, no human standing behind it.
    • Consequential specificity from the mouth of the persona — named tickers as instructions, not as labels on a diagram.

    What does not automatically fail: covering the topic; animating a public chart you actually exported; a founder on camera; a narrator who reports a filing with the filing on screen. The policy is the persona-as-adviser format. The topic is still legal to explain.

    Disclosure that the presenter is AI does not move a diagnosing host out of the bucket. Change the format.

    Diagrams are stills. Type is a stills row.

    Numbers a video model draws inside the frame have to be right every frame. That is a bad place to typeset supply, vesting, or a fee. Ideogram V4 is the stills row for posters, logos, and readable lettering: text-to-image, 6 credits, no duration, aspects 1:1, 16:9, 9:16, 4:3, 3:4. Spend the six credits. Read every glyph. Lock the PNG. Then move it, or hold it as a card.

    Do not prompt a video model to "animate a tokenomics pie that says 30% community." You will watch a cousin pie restype itself. Generate the diagram as a still. If it must move, image-to-video from that PNG with a motion line only — camera, not a second typesetting pass.

    Captions are a timeline job. If the founder spoke, burn transcribed captions. If the line is a disclaimer you wrote ("Educational. Not financial advice."), that is an authored overlay, not a transcript. Do not let the model invent the disclaimer inside the scene.

    Founder footage you actually have

    The trust signal in this category is a real mouth on a real take: AMA cut, desk recap, screen-share of a dashboard you operate. Generation can do b-roll, title cards, and assembly. It cannot be the adviser.

    Put the founder (or a named, credentialed person) on camera when the video tells a viewer what to do — or do not tell them what to do. Third-person description of how a mechanic works, with the source on screen, is a different object from "here's what you should ape." Keep the grammar in the first column.

    If you do not have founder footage, you do not get to generate a substitute host. You get diagrams, sourced cards, and a narrator that never recommends. That is a smaller channel. It is a channel that can still apply to YPP.

    FAQ

    Can an AI avatar narrate a tokenomics explainer?

    As a narrator over sourced diagrams, with no advice in the second person, it is not the Help example. As a "market host" telling viewers what to buy, it is. If you cannot tell which one you made in the first thirty seconds, rewrite.

    Why Ideogram instead of generating the chart in video?

    Because lettering is a stills problem. V4's catalog job is accurate text rendering on posters and logos at 6 credits. A video generate that restypes a supply number is a failed explainer, not a style.

    Does labelling the video AI save a synthetic analyst?

    No. YouTube's third bucket is the persona-plus-advice pattern, not the upload toggle. Change the presenter, or change the grammar.

    What if the founder will not appear on camera?

    Then the channel does not get an invented desk talent. Ship diagrams, captions, and attributed cards. Skip the market host.