Synthesia's new how-to guide says better videos from an AI video agent depend on five habits: setting the goal, audience and scope before prompting; supplying existing source material; naming the viewer and desired action; describing style plainly; and revising the first version as a draft through conversation.
Anyone can now describe a video in a sentence and get a finished one back from an AI video agent. Getting a good one still depends on two things: knowing how to ask for it, and having an idea worth asking for. That is the core argument of a new how-to guide published by Synthesia, the company behind the Syren AI video generator, which it describes as an AI video generator for business.
The guide is aimed at small businesses and small teams, which it says no longer need an editor, a designer or a big budget to make video. What they need, the guide says, is a clear message and a few working habits. It lays out five principles that make the biggest difference to what comes back after a prompt.
What should you set before prompting an AI video agent?
The first is to set the dials before typing anything. That means defining the goal, the audience and the scope. The goal is what the video should achieve. The audience is who will watch and what they already know. The scope covers how long the video runs, how many topics it touches and how long each scene lasts. The guide recommends a story arc with a problem, a tension and a resolution over a dump of information. Prompts should be written conversationally, the way a person might describe a mood board in words.
What source material should you give an AI video agent?
The second principle is to supply existing material rather than letting the agent guess. Useful sources include a help-center article, a PDF manual, a slide deck, a product page or a launch blog post. When the agent has access to brand guidelines, terminology and product details, the guide says, users spend less time fixing facts after the video is made.
Third, name the viewer and the action you want them to take. A video for a new hire and one for a prospective customer should look and sound different even if the facts are the same. The guide says the next step should be stated plainly, whether it is a sign-up, a booked call, a set of steps or a single thing to remember. It also varies the brief by video type. Brand films should prioritize emotion and originality over completeness. Product explainers should begin with the problem and present features as the way to solve it. Internal communications should be clear and honest, without spin. Social videos succeed or fail in the first second, so the prompt should say so.
The fourth principle concerns style. The guide advises against design jargon and recommends plain descriptions such as clean and calm, with lots of white space, or upbeat and fast-paced, like a product launch. Users who cannot put the look into words can upload images that match what they want. Brand colors, a logo or past videos the user likes can also help. The guide argues that business video has to look like the business that makes it, not like a generic template, and that an agent can only match a style it has seen.
How should you revise the first version of an AI video?
The fifth principle is to treat the first version as a draft. The guide says the main advantage of an agent is that users can revise through conversation. Requests like “Make the opening shorter” or “Make the tone warmer” work better than a general instruction to improve the video. Changing one or two things at a time makes it easier to see what each edit did, and the guide suggests stopping once the video says what the user intended.
The guide includes three sample prompts written by its team. One calls for a 45-second product launch video aimed at a specific audience, with a single call to action to book a demo. A second requests a 90-second all-hands update built from an attached memo, covering only three points. A third asks for a 20-second vertical teaser for a webinar, with a hook in the first second and the words “Register free” at the end. Each prompt contains bracketed placeholders for the user's own details.
For readers unsure where to begin, the guide lists five videos that small teams commonly plan to make: a 60-second explainer of what the company does, a how-to answering the questions customers ask most, an onboarding walkthrough for new customers or hires, an announcement for a new feature, offer or service, and a video version of a popular blog post or FAQ page.
The guide closes by pointing readers to Syren, which it says works like a video agent. Users start from a prompt or their own documents, the tool builds the video with the user's brand, and the user refines it through chat. Syren runs as a web app and inside ChatGPT and Claude. The guide's final point is that video is becoming a basic workplace skill, similar to writing a clear email or building a slide deck, and that the people who do it well know what they want to say and what they want viewers to do afterward.
Dieser Artikel wurde mit Unterstützung von KI verfasst.
News Factory APP - agentische News für besseres SEO & AEO.
