Google dropped a lot of announcements at I/O 2026 yesterday. If you're running a small or mid-size business, most of the keynote probably felt like it was aimed at developers and Silicon Valley types. Fair enough. But buried in the noise is something worth paying attention to: Gemini Omni.
I want to cut through the marketing language and talk about what this thing actually does, where it falls short today, and how businesses like the ones we work with at OrionX could put it to use.
So What Is Gemini Omni?
Gemini Omni is a new multimodal AI model from Google. "Multimodal" means it can take in text, images, audio, and video all at once, and produce output across those same formats. The headline capability right now is video. You give it a mix of inputs (a photo of your product, a text description of what you want, maybe an existing clip) and it generates a short video.
Google's own description: it can "create anything from any input." That's the marketing pitch. The reality is more specific. Right now, Omni generates short-form video clips (roughly 5 to 15 seconds), and you can edit them through conversation. Tell it to change the background, swap a character, adjust the camera angle, and it builds on what you said before. Each instruction stacks on the last one.
The physics simulation is notably better than previous AI video tools. Gravity, motion, fluid dynamics all behave more realistically. That matters because it's the difference between a video that looks obviously AI-generated and one that might actually pass as footage.
Where you can use it today: Google AI Plus, Pro, and Ultra subscribers through the Gemini app and Google Flow. YouTube Shorts integration is coming this week. Enterprise API access is not available yet, which limits what businesses can do programmatically.
What it costs: The entry-level AI Plus plan runs $20/month per user. Google also announced a new AI Ultra plan at $100/month that bundles priority access to Omni, higher usage limits, and Antigravity (their new coding platform).
Why This Matters If You Run a Business That Isn't a Tech Company
Most small businesses spend money on content they wish they didn't have to. Social media posts, product videos, marketing materials, explainer clips. The production cycle for a 30-second product video typically involves a videographer, an editor, maybe stock footage licences, and a week of back-and-forth. TechBullion estimated that the average small business spends between $2,000 and $10,000 per month on video content alone.
Gemini Omni compresses that. Not all of it, not perfectly, but meaningfully. A business owner could photograph their product, type a description of the ad they want, and have a usable short-form clip in minutes. Iterate on it through conversation. No editing software, no stock footage subscriptions, no freelancer invoices.
That's not theoretical. Google specifically demonstrated product reveals, explainer content, and localised creative during the keynote. VentureBeat noted that the practical enterprise use cases go beyond marketing videos into areas like training materials, internal communications, and product documentation.
The Other Announcement That Matters More: Gemini Spark
Omni gets the headlines, but Gemini Spark might be the more useful tool for day-to-day business operations.
Spark is a personal AI agent that connects directly to Google Workspace: Gmail, Docs, Sheets, Slides, Calendar. It runs on Google Cloud, which means it keeps working after you close your laptop. Google's VP of the Gemini App, Josh Woodward, said during the keynote that small businesses are already using Spark to monitor inboxes so they never miss a customer question.
In the demos, Google showed Spark pulling information from emails, docs, and sheets to draft status update emails. It synthesised meeting notes scattered across multiple apps into a single document. It monitored inboxes for specific triggers (customer questions, invoice requests) and flagged or responded automatically. One example had it reviewing monthly credit card statements for hidden subscription charges. The throughline is chaining together steps across Workspace apps that a human would normally do manually, one tab at a time.
The MCP (Model Context Protocol) integrations are worth watching too. Google announced connections with Canva, OpenTable, Instacart, and others. Over time, Spark will be able to take actions across third-party services directly.
Spark is rolling out first to AI Ultra subscribers in the US, with broader availability coming this southern hemisphere winter.
Where We See Actual Value for SMBs
We work with accounting firms, SMSF administrators, and professional services businesses across Australia. Here's where these tools create real value, and where they don't.
Content production gets cheaper. If your firm needs regular LinkedIn posts, client-facing video updates, or explainer content, Omni cuts the cost of producing that material. You're not replacing a full creative team. You're eliminating the $500-per-clip freelancer for routine social content.
Admin work shrinks. Spark's Workspace integration addresses the exact problem we hear about constantly: staff spending hours on email triage, meeting note compilation, and status reporting. For a 10-person accounting firm during tax season, having an agent that automatically compiles client correspondence and flags unanswered queries could reclaim hours every week.
Client communication gets faster. Generating a short video walkthrough of a tax return summary, or producing a branded explainer for new SMSF regulations, becomes a 15-minute task instead of something you outsource or skip entirely.
Internal training becomes feasible. Most SMBs don't produce training materials because the production cost doesn't justify it. When you can generate a training video from a text document and a few screenshots, the economics change.
What's Not Ready Yet
Omni doesn't have an API yet. That means you can't integrate it into your own systems or automate video generation programmatically. For businesses that want to build AI into their workflows at a systems level, you're stuck with the consumer-facing Gemini app for now.
Content restrictions are strict. VentureBeat reported that early testers found the moderation filters quite aggressive, which could limit legitimate business use cases. If you're in an industry that deals with sensitive topics (medical, legal, financial services), you may run into walls.
Spark is US-only at launch and limited to Ultra subscribers. Australian businesses will need to wait for broader rollout. And the Workspace integration, while promising, is still early. Google announced plans for browser control, custom sub-agents, and direct emailing capabilities, but those aren't shipping yet.
The video output is short-form: 5 to 15 seconds. That's fine for social media and quick explainers. It's not replacing your company's brand video or a detailed training walkthrough. Not yet.
One more thing: Google hasn't published independent benchmarks for Omni. The demos looked impressive, but demos always do.
What We'd Recommend
If you're on Google Workspace already (and most of the businesses we work with are), keep an eye on the Spark rollout schedule for Australia. The Workspace integration alone could justify the subscription cost for firms drowning in email.
For video content, the AI Plus plan at $20/month is a low-risk way to experiment. Try generating a few LinkedIn video posts or short product clips. See if the quality meets your bar. If you're producing more than a handful of videos monthly, the cost savings over freelancers will be obvious quickly.
Don't wait for the API if you need content now, but do wait for the API before planning any kind of automated pipeline. Consumer tools are fine for hands-on use. They're not a foundation for systematic business processes.
For more on building a practical AI investment strategy, see AI ROI for Small Business: What to Expect in Year One. And if you're trying to work out whether your current tech partner can actually help you adopt tools like this, our guide to choosing between an MSP and an AI integrator is worth a read.
