At Kaizen AI, we specialize in delivering innovative solutions that drive sustainable growth and success for your business, Let us help you transform your vision

Get In Touch

Google Gemini Omni: How to Make and Edit Marketing Videos Just by Talking

  • Home
  • Blog
  • Google Gemini Omni: How to Make and Edit Marketing Videos Just by Talking
Flat vector illustration of conversational AI video editing with a chat bubble and split before/after video frame

Google has quietly turned video editing into a conversation. Gemini Omni, launched at Google I/O 2026 and opened up to developers on 30 June, lets you generate a short video from a text prompt, then keep talking to it: “change the background to a Scandinavian kitchen”, “make it golden hour”, “add a glass of water next to the product.” Each instruction builds on the last one without losing the camera angle, the product, or the audio. No re-rolling from scratch, no starting the brief again.

For a UK small business trying to keep marketing video in-house, that changes the maths on what’s worth attempting yourself and what still needs a professional.

What Gemini Omni actually does

Gemini Omni Flash is Google’s fast, cost-efficient model for generating and conversationally editing short video from text, images, audio and existing video references. It produces clips up to ten seconds long with synchronised audio, and it is available three ways: inside the Gemini app for everyday use, through Google Flow for more controlled creative work, and via the Gemini API in Google AI Studio for anyone building it into their own tools.

The part worth paying attention to is the editing loop. Most AI video tools give you one shot: type a prompt, get a clip, and if it is wrong you write a new prompt and hope the next version is closer. Gemini Omni keeps session history through its Interactions API, so a follow-up instruction is interpreted against everything you have already built, not treated as a fresh request. That is the difference between briefing a video and re-briefing it five times.

What it costs, and what that replaces

Through the API, Gemini Omni Flash charges $0.10 per second of output, so a ten-second clip costs roughly a dollar, about 80p. The consumer app runs on a credit system instead: Google’s AI Plus plan is $4.99 a month for 200 credits, Pro is $19.99 for 1,000, and Ultra runs up to $199.99 for 25,000, all usable across Omni video generation and Google’s other creative features.

Compare that with hiring it out. The average day rate for a freelance video editor in the UK sits around £350, with corporate and commercial work often running £300 to £800 a day depending on experience and finish. That is not a reason to stop using editors, a polished brand film still needs a human eye, but for the quick social clips, product teasers and ad variations that used to eat a chunk of a marketing budget, the economics have moved a long way in a short time.

The UK adoption picture, and the catch

AI use among UK SMEs has climbed fast: 54% now report actively using AI, according to the British Chambers of Commerce’s 2026 survey with Atos, up from 35% in 2025, 25% in 2024 and 23% in 2023. Marketing is consistently the most common use case cited.

Here is the uncomfortable part. Adoption is not the same as return. Across the same wave of UK surveys, the vast majority of businesses using AI report productivity gains, but only 12% say it has actually increased revenue so far. Buying access to a tool and turning it into a result are two different projects, and the gap between them is where most small businesses lose momentum.

How we’re using it

We build the marketing automations behind this blog ourselves, so we have been testing Omni-style conversational editing directly against the kind of short-form video clients ask us for: product clips, service explainers, the kind of thing that used to mean three rounds of email with a freelancer before you saw a version worth posting. The conversational step is the real time-saver. Rather than exporting a clip, opening it in a separate tool and re-cutting it, you keep refining it in the same thread until it is right, then export once.

It will not replace a videographer for anything that needs a real location, a real product in someone’s hands, or a brand film that has to carry a campaign for months. What it is good for is the volume work: the five versions of a fifteen-second ad you need to test which hook actually lands, or the quick clip a service business wants for a single week’s promotion. That is exactly the kind of output that used to get skipped because a proper edit cost more than the campaign was worth.

Getting started without wasting a month

If you want to try this properly rather than poking at it once and forgetting it exists, start narrow. Pick one short, low-stakes video you already need this month, a product teaser, a testimonial cut-down, a single ad variation, and run it through the conversational loop end to end before judging the tool. Note how many turns it actually took to get a usable result, because that number, not the headline price, is what tells you whether it fits your workflow.

Then test whether the output holds up on the platform you are posting it to. A clip that looks sharp in the preview window does not always survive compression on Instagram or TikTok, so check the real upload before you build a whole campaign around it.

The tools keep getting cheaper and faster. The businesses that get ahead are not the ones who adopt first, they are the ones who work out, quickly and cheaply, which jobs a tool like this is actually good for before they commit real budget to it.

Leave A Comment

Fields (*) Mark are Required