When a Marketing Team Can Ship Five Videos a Week

A while back, a friend of mine who runs a consumer brand complained to me.

He said, "Liu Run, my biggest pain right now isn't lack of traffic, and it isn't lack of budget. It's the design schedule. A single short video — from idea to finished cut — his design team grinds on it for a week. One week, one video."

I said, "Well, have you ever thought that this whole pipeline can actually be taken apart?"

He froze for a second.

That's what I want to talk about today. Something called AIGC is quietly transforming marketing content production — from an artisan workshop into an assembly line.

What Exactly Is AIGC?

It's simple.

AIGC stands for AI-Generated Content. Text, images, audio, video — all the things that used to require someone to draw stroke by stroke, frame by frame. Now, generative AI handles the first draft for you.

It's not about helping you look things up or run analysis. It's about directly producing something new.

That's the biggest difference between it and traditional software. Traditional software "understands what you give it." Generative AI "creates something new for you."

Think about it — what does that mean?

It means a small team can now produce the kind of visual volume that, in the past, only big ad agencies could afford. The playing field has been leveled.

The Real Bottleneck Isn't the Tools — It's the Blank Page

I've talked to quite a few marketing leaders, and where do they get stuck? On that blank page.

Every time a new brief comes in, the designer has to gather references, build a mood board, sketch out concepts — just these prep tasks eat up a day or two. By the time they actually start on the real work, there's barely any time left.

What AIGC eliminates is exactly this stretch.

You no longer have to stare at a blank page. You can have it give you dozens of directions in a matter of seconds. Then all your energy goes into the few worth keeping — picking, refining, polishing.

But there's a misconception I have to mention here.

A lot of people treat AI as the last step in the process: "Once we're done, let's just run it through AI to spit out a couple of alternatives." Wrong. It's not a finishing tool — it's an engine that runs through the entire process.

The people who really know how to use it embed AI into every single stage:

• Breaking down the brief, finding directions, building a shot list — let AI run a first pass;
• Generating the first draft of images and video, picking the most suitable model for each shot;
• Recomposing, resizing, removing backgrounds, adding subtitles — all done on a single canvas;
• Exporting in every aspect ratio, all in one go.

When all four steps live in the same place, you realize that "one video a week" can become "five videos a week" — and the brand feel doesn't drop.

Will AI Steal Creative People's Jobs?

That's the question I get asked the most.

My answer is: No. But it will redistribute the work.

AI is powerful — it's fast and generates tons of directions. But it's missing one thing: empathy. It doesn't understand why your consumer gets emotional late at night. It doesn't know what a certain color combination means in your category. It has never lived.

So the smartest way to use it is to treat AI as a junior teammate who's lightning-fast but still needs someone to guide them.

How do you guide them?

Teach it your brand, once.

Feed it your brand guidelines, color palettes, reference images — all at once — and let it memorize them. That way, every time it produces something, the default is your brand's tone, rather than drifting wherever the wind takes it. On Melius, this has been built into a "reusable skill" — you save it once, and every subsequent generation carries your brand DNA.

Then, humans step back into two roles: editor, and strategist.

Emotion, fact-checking, that final decisive judgment — leave those to people. Hand the exhausting, repetitive grunt work of execution over to AI.

That's a healthy division of labor. Let AI do the heavy lifting; let humans bring the flavor.

So, How Does Google View AI-Generated Content?

My friends who do SEO care about this the most. After all, search engines used to penalize machine-generated spam.

I looked into it, and Google's stance is actually very clear: it rewards quality, not provenance.

What it looks at is the E-E-A-T framework — Experience, Expertise, Authoritativeness, Trustworthiness. Your content, whether written by a human or generated by AI, as long as it genuinely answers the user's search intent and reads naturally, it'll rank.

So don't waste your energy on "evading AI detection." That's a false premise.

Your energy should go into making your content substantive. Answer why the user searched for that term, cover all the related sub-questions around the topic, and establish your authority in a given field. The same goes for images — original, brand-aligned visuals do a far better job of retaining readers than generic stock library photos.

How Do You Choose a Tool? Look at Four Things

There are hundreds of tools on the market claiming to do AIGC. How do you pick?

My advice: don't watch the demos. Look at these four things:

First, model coverage. No single model is good at everything. Product photography, cinematic shots, text-embedded graphics, fast turnaround — each has its own strengths. So a platform that can orchestrate many models will always beat one locked into a single vendor. You need to be able to switch models depending on the job.

Second, collaboration. Generating, editing, approving — these things should ideally happen on the same canvas, in real time. No files flying back and forth over email, sending a new version with every revision, until nobody can keep the versions straight.

Third, is there an agent that can actually get work done? Note — not an assistant that gives one answer per prompt, but an executor that can break down tasks on its own and run multi-step workflows. You give it a brief, and it plans on its own, picks the right models, and runs the entire pipeline end to end.

Fourth, is the pricing transparent? One subscription with all models sharing a single quota pool is far more cost-effective than subscribing to a separate plan for every model you turn on. You need to be able to do the math clearly.

Melius has nailed all of these. All mainstream image, video, and audio models share a single subscription; the canvas supports real-time multi-person collaboration; and that agent called Mel can set up and run the entire workflow for you.

Put bluntly, that fragmented workflow where you used to jump between five or six tools has now collapsed onto a single canvas.

Three Things Most Worth Doing

Once you've picked your tool, how do you squeeze every drop of value out of it? I've distilled three habits.

One, write your prompt like a "shot specification."

"Make me an ad" — that kind of instruction only gets you mediocre results.

A strong visual prompt should read like you're briefing a cinematographer: what's the subject, what's the composition, where's the light coming from, what lens, what mood. For example: "Top-down shot, matte black skincare bottle, placed on a wet stone slab, soft light entering from the left window, shallow depth of field, editorial beauty style." See — that's much more specific.

Take it a step further: hand the entire brief directly to Mel, and it can translate your intent into a pre-wired, multi-step workflow on its own.

Two, turn one asset into countless assets.

Mass marketing is outdated. What people compete on now is hyper-personalization. A single concept — you need to be able to slice it into dozens of versions: different channels, different audiences, different tests.

This is something AI is naturally good at. One design, and in seconds it rearranges every aspect ratio you need for social media, web, and print. The cost of adapting for each format goes from "reshoot" to "a few minutes."

Three, keep your eye on ROI.

You adopted this tool to cut costs and boost efficiency. So go measure it.

Figure out exactly how many hours your team saved on drafts and production, then redirect that time into high-value work — like strategy, distribution, building community. Production costs go down, output and quality go up — and those returns compound.

In Closing

That friend of mine who complained about the design schedule eventually went and tried it.

Two weeks later, he messaged me with just one line: "Liu Run, this week we shipped seven."

I didn't reply. But honestly, I was happy for him.

AIGC has become the new foundation of digital content production. The brands that will dominate feeds and search results in the years ahead are the ones that take AI seriously, teach it to understand their brand, and consistently produce genuine value.

Put generation, editing, assets, and collaboration all onto a single canvas. Then, spend the energy you've saved where humans should be spending it most.

That, I'd say, is the biggest dividing line among marketers in the years to come.