The same source material — an article, a transcript of someone else's video, or a manually written text — can be transformed into three completely different videos depending on how AI handles the text when preparing the voiceover. This is why at Telematic, the video script is formed not in one universal way, but in three different modes, and choosing a mode is essentially the first and most important decision when working with the material, even before you start thinking about visuals or voice.
The three modes — full text, hybrid, and teaser — serve different purposes and handle the source facts differently. The difference between them is not cosmetic: they use different text processing algorithms, and the choice of mode affects how literal the video will be, how lively the speech will sound, and what will actually appear on screen.
Full Text: Extractive Accuracy
The "full text" mode works literally — extractively. The AI does not paraphrase or embellish; it takes the source material and turns it into voiceover while preserving the wording, numbers, and details as much as possible. This mode is suitable for situations where citation accuracy is crucial: analyses of regulatory documents, financial analytics, technical instructions, news with specific figures and dates.
The main advantage of this mode is predictability. You know exactly that nothing will appear in the video that wasn't in the source, and no important fact from the article or transcript will be lost. The downside is that such text does not always sound like natural speech: if the source is written in bureaucratic language, the extractive script will inherit that dryness.
Hybrid: Lively Storytelling Over Facts
The "hybrid" mode is structured differently: the AI does not just extract text; it unfolds the source into a full-fledged voiceover, adding transitions, intonational shifts, and a more conversational delivery — while preserving the facts, numbers, and details of the material. Essentially, this is a reworking "inspired by" the source in a good sense: the voiceover sounds like a person telling a story, rather than a narrator reading a paragraph from an article.
This mode works particularly well when the source is a dry corporate text, an industry news piece, or someone else's transcript with verbose phrasing, and you need an engaging video for social media or YouTube. The AI takes on the role of the editor: it retains the essence but changes the form of delivery to suit natural speech.
How to Choose Between Full Text and Hybrid
The practical rule is simple: if the video needs to quote the source as accurately as possible — choose full text. If the video needs to tell a story inspired by the source and keep the viewer's attention — choose hybrid. Many teams develop a habit over time: news and analytical formats go through full text, while educational and engaging videos go through hybrid.
Teaser: Short Trailer Without a Call to Action
The third mode — "teaser" — does not work directly with the source material but with the script of the already prepared main video. A teaser is a short vertical trailer that announces a horizontal video: it takes key moments from the script, compresses them into a dynamic short format, and presents them as a hook.
An important feature of the teaser is that it fundamentally lacks a call to action. The format is deliberately limited: it is not a separate standalone video with a funnel, but an announcement, whose sole purpose is to pique the viewer's interest and encourage them to watch the full video. While in other formats, the CTA funnel is subtly woven into the text and clearly stated in the description or the first comment, in the teaser, this logic is intentionally disabled — otherwise, the announcement function mixes with the sales function.
The teaser is usually generated after the main video script is ready and is automatically linked to it — thus, the output results in a pair of "horizontal video + vertical announcement," which can be conveniently published simultaneously on different platforms.
How the Script Mode Combines with Other Settings
All three modes work in conjunction with two additional adjustments — the script volume slider and the narrative pace. The volume slider compresses or expands the material: in "full text" mode, it extractively selects part of the source, while in "hybrid," it manages the degree of compression during the AI reworking. The narrative pace, in turn, determines how the phrases are restructured — dynamically, standardly, or calmly — regardless of which script mode is chosen.
This means that the mode is not the only variable. For example, a long analytical article can be turned into both a detailed ten-minute video in full text and a compressed dynamic video in hybrid — while both options remain true to the facts of the source, just with different degrees of literary processing.
Our Experience Transitioning to Different Modes
When we transitioned our corporate blog to video format, the first attempt was full text — intuitively, it seemed that since the material was already well-written, why "retell" it. In practice, it turned out that written and spoken text are different genres: phrases that read well on the page sound cumbersome when voiced. After transitioning some videos to hybrid, audience engagement noticeably increased — not because the facts changed, but because the delivery became closer to conversational speech.
In contrast, one of our clients, who creates detailed analyses for a B2B audience, faced the opposite situation: the audience expects precise quotes and figures, not a "story inspired by" the source. For this format, hybrid proved to be excessive — the lively delivery was perceived as unnecessary dramatization of a serious topic, and the team stuck with full text, only adding teasers for social media announcements.
Practical Cheat Sheet for Choosing a Mode
If the task is to quote the source as accurately as possible (regulatory, finance, technical documentation), choose full text. If the task is to tell an engaging story inspired by dry or foreign material, choose hybrid. If you already have the main video ready and need a short announcement for vertical platforms — generate a teaser based on it, keeping in mind that it will not contain a direct call to action.
There is no "correct" mode for all cases — there is a suitable mode for a specific task, audience, and platform. A sensible strategy is not to commit to one mode forever, but to test both on the same material and see how your audience responds.