When the numbers have to be right, do not generate pixels

The problem with generative video for data

A generative video model produces frames. It does not know that 4.25 percent is a rate and not a decoration. Text rendered inside generated frames can come out distorted, and a number can change from one shot to the next. For entertainment that is a quirk. For a markets note, an earnings summary or a research report it is a correction waiting to happen.

Publishers in finance and data journalism need the opposite property: whatever is on screen should be the text and figures from the source, drawn cleanly, the same way every time.

What a template based renderer does differently

Blog2Video does not generate footage of your article. It reads the article, splits it into scenes, and renders each scene in a designed template with real text, real layouts and chart scenes built from the figures in the source. The same template produces the same look on the first video and the five hundredth.

• On screen text is rendered as text, not drawn as an image.
• Templates keep fonts, colors and layout consistent across every video.
• Chart scenes are built from the figures in your content.
• A custom template carries your brand into every output.

Put a human gate before publish

The narration script is written by an AI model, so treat the first draft like a junior analyst's draft: useful, fast, and not yet signed off. The project API has a script review stage. Create a project with script_review_enabled and generation pauses after the script is written, so an editor can read it, fix any figure, and approve it before the video renders.

That one step is what makes the workflow safe for regulated or reputation sensitive content. Nothing reaches a render, and nothing reaches your audience, without a person approving the words. The template guarantees how it looks. Your review guarantees what it says.

A workflow for a research or markets desk

Here is one pipeline for a research or markets desk. It uses only calls that exist today.

• Publish the note or report in your CMS as usual.
• A job sends the article text or its URL to the API and creates a video with your custom template.
• Script review holds the project until an editor approves the script and checks the figures.
• The render runs, your job polls until it reports done, and stores the MP4 from video_url.
• The video is attached to the original article, or posted to YouTube, LinkedIn and your newsletter.

Why this is a platform opportunity too

If you build tools for finance media, a newsletter platform, a research portal or a CMS, you can offer video as a feature without building a renderer. Create each user's videos through the API, tag them with an external_user_id, and keep your own ownership checks in front of every request. Your customers get video, you keep one integration to maintain.

See the full API overview

What this does not do

Be clear about the limits when you evaluate it. There are no AI avatars, no prompt generated footage and no webhooks, so you poll for status. Output wording can differ between runs, which is why script review matters. Paid plans start at 9.99 dollars a month, and API access needs a paid plan.

Read the API documentation

Frequently Asked Questions