How to make a social video with Gemini, step by step (the hard way)

Of the big chat tools, Gemini has the most video built in. It writes, it makes images, and on paid plans it makes video clips in the chat. Google's newest model, Gemini Omni, launched on May 19, 2026, and makes and edits video by conversation.

Clips are not a finished social video, though. Below is the full route, from the script to a scheduled post. It works. It is the hard way, and you should know what it takes.

Prices are monthly billing, read on October 28, 2026, from each tool's own page unless a line says otherwise. They change often. The times are rough estimates for someone who knows the apps. Your first video will take longer. For ChatGPT or Claude, see the ChatGPT guide and the Claude guide.

Step 1: Write the script and the scene list in Gemini

Cost: Google AI Pro is $19.99 a month. AI Plus, at $4.99, now lists video in the Gemini app too, with fewer Flow credits. Time: about 15–30 minutes.

Ask for the script and the pictures at once. For example: “Write a 30-second voiceover for a vertical video about our spring sale at a garden shop. Hook in the first line. Short sentences. One call to action. Then split it into five scenes. For each scene, write a video prompt: the subject, the setting, the camera move and the light. No words in any scene.”

Read the script aloud with a timer. Each scene should match one or two lines of voice. Ask Gemini to tighten anything that runs long.

Step 2: Generate the clips in the Gemini app

Cost: included in your plan. Video in the Gemini app needs a paid plan. AI Pro includes a trial of Veo 3.1 Lite; AI Ultra, from $99.99, has full Veo 3.1. Time: about 30–60 minutes, most of it regenerating.

Paste the first scene prompt and ask for a vertical clip. Watch it. If the motion is wrong, say what to change in plain words: “slower camera, less sky, keep the pots in frame.” Omni edits by conversation, so you do not have to start over each time. Keep the best take, download it, and move to the next scene.

Two cautions. Generated video carries a SynthID watermark, Google's marker for AI-made content. And some reviewers report spelling errors in the text of generated videos, so keep words out of your clips. Add text in the edit instead.

Step 3: Build matching scenes in Flow

Cost: included. Flow credits come with each plan: 200 on AI Plus, 1,000 on AI Pro, with Omni Flash. Time: about 30–60 minutes. This step is optional.

The Gemini app is quick. Flow, Google's filmmaking tool, came to Workspace in January 2026, and it is better when your scenes must look alike. Following Google's Flow help: open a project and describe the scene. Click the model name and choose Video. Under Ingredients, add reference images, such as your product or your shop, so they show up in every clip. Under Frames, add a start or end image and describe the move between them. Set the aspect ratio to vertical, choose how many versions you want, and generate.

Use your own photos as ingredients. A clip of your real storefront beats a made-up one. Download the clips you keep.

Step 4: Edit and add the voice in CapCut

Cost: CapCut Pro is $19.99 a month on monthly billing; a no-cost tier has the basics. Time: about 45 minutes to 1½ hours.

Start a 9:16 project. Record the voiceover first, straight into CapCut on your phone, or import one from a voice tool such as ElevenLabs. That would be another bill. Lay the clips over the voice in order, then trim each one so the cut lands on a word. Generated clips often start or end with odd motion, so cut the first and last half-second. Add music well under the voice and check that the track is cleared for business use.

Step 5: Add captions and on-screen text

Cost: included in CapCut. Time: about 15–30 minutes.

Captions let the video work with the sound off. In CapCut, generate auto captions from the voice track. Then read every line. Auto captions mishear names, prices and street names. Fix them by hand. Add the on-screen text from step 1 as separate text layers, kept clear of the bottom of the frame where each app puts its buttons. Export at 1080 x 1920. Export again for any other size you post.

Step 6: Make the cover in Canva and schedule it in Buffer

Cost: Canva Pro is about $18 a month (as reported by Piktochart; Canva's own page could not be read). Buffer Essentials is $5 a month per connected account, so Instagram, TikTok and Facebook come to $15. Time: about 25–35 minutes.

In Canva, make a 1080 x 1920 design. Drop in your best frame and a short title in big type, kept to the middle so profile grids do not crop it. Download it.

Gemini's plans list no posting or scheduling to your social accounts. So open Buffer, create a post, upload the video and the cover, paste the caption, adjust it for each platform, pick a time and schedule it. Then start the next video.

The honest tally

Time: add up the estimates and one 30-second video takes roughly 2¾ to 5 hours once you know the apps. Skip Flow and it is a little less. Regenerating clips and the edit are the biggest blocks.

Money: Google AI Pro $19.99, CapCut Pro $19.99, Canva Pro $18 and Buffer at $15 come to about $73 a month for four subscriptions. That is the cheapest of the common chat stacks; Claude's is about $103.

And you are the glue. You write the prompts, regenerate the clips that came out wrong, edit, caption, design the cover, check each post and remember to post. Gemini makes beautiful seconds of video. It does not plan your week. How the three chat tools compare has the full picture.

In dollars, that stack costs less than VEXI. In hours, it costs you the afternoon.

Or just ask VEXI

VEXI costs more than the stack: $199, $399 or $599 a month, and every plan starts with a 15-day trial. What you pay for is the work you no longer do.

One chat does the whole job. Type what you want the way you would say it to a person. It writes the script. Videos come back cut, with B-roll and captions, sized for each platform, without you opening an editor. Send a photo and it becomes a moving clip, sized for the feed. Upload a long raw recording and get back the clips that work best, captioned, with music under each cut.

To change something, tap it on the design or say what you wanted instead. Then it posts for you to Instagram, Facebook, TikTok, YouTube, LinkedIn, Pinterest, Threads and your Google Business Profile. Autopilot writes, designs and posts your week ahead, at the times your audience shows up, and it can wait for your approval on everything.

So yes, you can do it the hard way. Or just ask VEXI.

See what VEXI makes, compare it with Gemini, or see the plans.