OneMob's AI video composer makes two kinds of video. With avatar puts an AI presenter on camera to deliver your message. Presentation builds a narrated motion video, with no presenter, branded with your organization's logo and colors. This article helps you pick the right one and write a prompt that gets a strong result on the first or second try.
The rule for both: be specific. The AI only knows what you tell it and what you attach.
Choose the right kind of video
Pick the mode by asking who should carry the message: a person or the content itself.
- With avatar works best when a face builds trust: a personal introduction, a follow-up after a call, a renewal check-in, an executive message to an account. The video is one presenter speaking to the viewer.
- Presentation works best when the content is the point: a product overview, a launch announcement, a results summary, a feature walkthrough, a training module, a short social post. The video is a sequence of branded scenes with on-screen text, images and narration.
If you are unsure, think about the first five seconds. If you picture someone saying hello, choose With avatar. If you picture a title and a headline number, choose Presentation.
Start a video
- Click + in the navigation menu and select Generate AI Video.
- In the Create video with AI window, choose With avatar or Presentation under the title.
- Type your prompt in the prompt box.
- Click + inside the prompt box to attach a script or documents from your Library.
- Set the options for the mode you chose. They are described below.
- Optionally, click the Improve prompt button, the sparkle icon next to +, to have OneMob rewrite your prompt with more detail.
- Click Generate.
OneMob confirms with AI video generation started and adds the video to your Library. While it generates, the Library item shows the current step. Generation can take several minutes. When it finishes, the item becomes a playable video like any other.
To see these tips in context while you write, click Best practices at the bottom of the composer.
Write a prompt for a With avatar video
The prompt box asks What should your avatar say? Answer that question. Write the message the presenter delivers, or describe it clearly enough that the AI can write it for you.
- Write for the ear. Short sentences, the way you would talk to a colleague. Read it aloud before you generate; if a line is awkward to say, it will sound awkward in the video.
- Lead with something real. A specific moment or number from your own experience beats a list of features.
- Add a tone line, not timestamps. End with how it should feel, such as "Tone: warm, confident, no corporate filler." Scene-by-scene timestamps produce a choppy result.
- Lock exact wording with a script. If the wording must not change, such as approved legal language or a set value proposition, click + and choose Add script.
- Choose the presenter. Click the avatar chip to use your own avatar or a stock avatar instead of Auto.
Example:
A 30-second follow-up to Dana, VP of Sales at Northwind, after today's demo. Thank her for the time, recap that she cared most about cutting rep ramp time, and say I will send the rollout plan by Friday. Close by asking for 20 minutes next week with her enablement lead. Tone: warm, direct, like a peer, not a vendor.
For more on scripts, avatars and translation-ready videos, see The Ultimate Guide to Writing Better Prompts for AI Video Generation.
Write a prompt for a Presentation video
The prompt box asks What's the video about, and who is it for? A Presentation video is built scene by scene from your description, so describe the content and the audience, and let OneMob handle the design. Your branding is applied for you, so you do not need to describe colors, fonts or layout.
A strong Presentation prompt covers five things:
- Topic and purpose: what the video is about and what it should achieve. For example, introduce our new analytics dashboard to existing customers.
- Audience: who watches it and what they already know. For example, operations managers who use the product daily.
- Key messages, in order: the three to five points the video must make, in the order you want them. Each point tends to become a scene.
- Proof: the facts, numbers or customer results to show, and which attached document they come from.
- Call to action: the one thing the viewer should do at the end.
Template to copy:
A [length] video about [topic] for [audience]. Goal: [what the viewer should think or do]. Make these points in order: [point 1], [point 2], [point 3]. Use the figures from the attached [document]. End with: "[call to action]." Tone: [tone].
Filled-in example:
A 1-minute video about our Q3 platform release for existing customers in operations roles. Goal: get them to try the new scheduling view this week. Make these points in order: scheduling now takes one screen instead of three, the new conflict alerts, and the time saved per week from the attached pilot results PDF. End with: "Open Scheduling and try it today." Tone: clear, confident, upbeat.
Tips that make the difference
- Give each scene one idea. Write your key messages as short, separate points. A paragraph that mixes five ideas produces crowded scenes.
- Put your numbers in a document. Numbers shown on screen come only from the documents you attach. If the video needs a number you did not provide, it is labeled Example data. Attach the source, such as a one-pager, a results summary or a pricing sheet, and point to it in your prompt.
- Attach the images you want to see. Images you attach appear in the video. Product screenshots and photos work best.
- Write the narration you want to hear. If a phrase matters, put it in quotation marks in your prompt.
- Match the length to the number of points. Two or three points fit in 30 seconds; four or five need 1 to 2 minutes.
Presentation options
- Voice: the narrator. Choose Sarah, Marcus or Ava. Sarah is the default.
- Music: background music under the narration. Choose No music, Calm or Upbeat. Music plays quieter than the voice and fades out at the end. The chip reads None when no music is selected.
- Aspect ratio: choose by where the video will play. 16:9 (the default) for microsites, email and presentations; 9:16 for mobile and stories; 1:1 for social feeds.
- Length: Auto length, 30 sec, 1 min or 2 min.
How your branding is applied
Presentation videos use your organization's logo and default theme color. If your organization has not set them, OneMob uses the logo and color it finds for your company's email domain, and otherwise a default blue with no logo. To get your exact brand in every video, ask your admin to set your organization's logo and theme color before you generate.
Use files from your Library
Files are the fastest way to make a video specific to your product, your customer and your results. Click + inside the prompt box and choose Add documents to pick files from your Library or upload new ones. Attached files show as chips that you can remove before you generate.
What to attach:
- For facts and numbers: a PDF one-pager, case study, results summary or datasheet. In a Presentation video, this is the only source for numbers on screen.
- For visuals: PNG or JPG product screenshots, product photos or diagrams. In a Presentation video they appear in scenes; in a With avatar video they give the AI visual context so it does not fall back on generic footage.
- For messaging: a saved script, through Add script, when a With avatar video must use your exact words.
A few habits help:
- Attach a few focused files rather than many. A two-page summary works better than a 40-page report, because the AI has less to sort through.
- Say how to use each file. For example, "Use the before-and-after numbers from the attached case study" or "Show the dashboard screenshot when introducing the new view."
- Keep your Library current. Replace outdated one-pagers and screenshots so the whole team generates from the latest material.
Revise a finished video
You do not need to start over when something is off. Open the video and use Edit video with AI to describe the change. Be as specific as you would be with a designer:
- Instead of: "Make it better."
- Try: "In the second scene, replace the headline with 'Scheduling in one screen' and show the dashboard screenshot from the attached file."
In a Presentation video, scenes you do not mention keep their content. Ask for one or two changes at a time so you can see what each one did.
Troubleshooting
- A Presentation scene shows "Example data". The number was not in your attached documents. Attach the document that has it and mention it in your prompt.
- An attached image did not appear. Say in your prompt where it belongs, such as "show the product screenshot when introducing the feature."
- The logo or colors are wrong. Ask your admin to set your organization's logo and theme color, then generate again.
- The scenes feel crowded. Cut to fewer key points, or choose a longer length.
- Text looks cramped on phones. Generate the video in 9:16 or 1:1 for mobile and social.
- The avatar delivery sounds robotic. Remove scene timestamps and write a flowing message with a tone line.
- Required wording changed. In a With avatar video, use Add script. In a Presentation video, put the exact phrase in quotation marks.
- The Library item shows "Generation failed". Read the reason shown on the item, adjust your prompt or files, and generate again.
Related articles
Was this helpful?