One marketer = team of 5 roles. I count by the hour and in rubles

On the last project, I filled five roles: videographer, editor, copywriter, voice actor and creative designer. Not out of love for experiments - out of arithmetic. Veo 3.1, Nano Banana Pro and ElevenLabs do in two hours what previously took two weeks. I count in this article in hours and in rubles, without rounding in my favor.

One thing that infuriates me is when “experts” talk about the AI revolution and don’t give a single figure of their own. Last week I came across another article: “Neural networks open new horizons for marketing.” Zero metrics. Zero cases. Zero rubles and hours. I’m the opposite: for the last year I’ve been analyzing all client budgets by hour and by unit-economics. This analysis comes from there.

AI 2026 is not the “next wave of tools.” This is a change in the exchange rate between the marketer’s time and the cost of production.

Previously: 5 people, 14 days, 65,000 ₽ per video

The canonical pipeline of a production team until 2024 looked like this. I write a brief, give it to the team, and two weeks later I receive a finished commercial. All this time - five people on three tasks, three synchronized calls, two rounds of edits and one explanation as to why the voice “doesn’t sound like the brand.”

Specifically in terms of roles and money - on one e-com electronics project in 2023:

• Copywriter — 8,000 ₽ per video. Brief, script, two edits.
• Videographer/actor — 25,000 ₽. Half-day shooting, editing not included.
• Editor — 15,000 ₽. Premiere Pro, color scheme, graphics.
• Voice actor — 12,000 ₽. Studio recording + retake.
• Creative designer (posters, banners) - 5,000 ₽ per set.

Total: 65,000 ₽ per video, cycle 14 days. I released 2 videos a week - that is, 130,000 rubles only for production. And this is without a budget for accommodation.

At the moment, no one was counting by the clock—everyone was counting by “pieces.” This is convenient for the client’s bills, but inconvenient for understanding where the leak is. I started counting.

Now: me + 3 tools, 2 hours, 8,000 ₽ per video

In 2024, I worked with MM AI Trading, a fintech project in the UAE, and introduced AI content production processes into the team (I am part of this stack I'm reviewing AI tools 2026). The summary noted a 40% reduction in content costs; the original team and cost structure requires reconciliation. What I collected:

Veo 3.1 – video + sync audio

Replaces videographer and editor. Prompt: “a marketer in headphones looks at the VK Ads dashboard, slow zoom-in, warm lighting, 8 seconds.” In 90 seconds - a finished clip in 1080p with synchronized ambient sound. Rate: $0.50 per second of video. Eight-second video - ~$4. Previously, one such shot cost 10,000 rubles for filming.

ElevenLabs v3 — voiceover

Replaces the voice actor. I clone a voice once (my own or with the client’s permission), then any text in 30 seconds. Subscription cost is $22/month unlimited. For 4 videos a week, that’s 4 rubles per video instead of 12,000 rubles for the actor. 3000 times cheaper. The quality in 2025 has ceased to be distinguishable from a live announcer - I checked this in a focus group test, no one guessed right.

Nano Banana Pro - posters and banners

Replaces creative designer. Generation of a poster with built-in Russian text - 8 ₽ per option. Fifty mesh options for 400 ₽. Previously, the designer made 5 layouts per day for 1,000 rubles. Now - 50 options in half an hour, and I can run them through an A/B test without having to negotiate with the designer about “one more color.”

What has been added - but this is not a “command”

I'm like a marketer. Brief, context, final taste. This is not “5 roles in one” - it’s one role that was previously blurred between six. Strategy + production curator, nothing more.

We calculate the final pipeline:

• $4 – Veo for an eight-second clip
• $1 – ElevenLabs for voice acting
• $3 – Nano Banana for visuals
• ~3,000 ₽ — my hour of work (brief + prompt + assembly)

Total: ~8,000 ₽ per video. Cycle - 2 hours. Compare with 65,000 ₽ and 14 days. According to unit-economics, this is 80 times faster and 8 times cheaper - and in fact, 240 times more efficient in terms of “person hours per unit of output.”

What has not been replaced - and will not be replaced in the next 2 years

I’m not writing that AI has “replaced everything.” If you remove me tomorrow, the pipeline will stop, because none of the tools themselves understand, why a video is being made. What is left behind a person:

1. Strategy and brief. What insight about the client is the basis of the video, what CTA, what funnel. AI does not know the context of your product - this needs to be put into the prompt, and before the prompt - into the head.
2. Final taste. Veo makes 4 variations of one prompt, and they all technically correct. But one of them “works”, and the other three look like stock material. This choice is made by a person.
3. Relationships with the client. AI won't call at 7:00 pm on Friday to explain why I'm not canceling the campaign because of one bad day on the dashboard. It's still about trust.

This three - strategy, taste, relationships - is “marketing 2026”. Everything else in the pipeline (production, retail, circulation of variations) is an operation that AI closes.

“But it doesn’t work in our niche” - and what the opponents are right about

The most common counterargument I hear is: “This works for you in performance, but for us the premium segment / people in the frame / subtle creativity.” They are right about one thing.

In b2c with an emotional product - where you need real faces, real situations, story-driven content - AI is still merging. Veo makes technically clean synthetic scenes, but there is no “life” in them. If I’m shooting a flagship video for Procter & Gamble, probably not through Veo.

But 95% of marketing production is not flagship. These are scripted videos for retargeting, posts with a product for social networks, animated teasers, voice-over for YouTube Shorts. AI closes all this already now — and over the next year it will catch up with the remaining 5%.

When I hear “it’s not working for us,” I usually ask one question: “Have you calculated in hours and rubles how much your current pipeline costs?” In 9 cases out of 10, no. Because it’s inconvenient to count: it becomes clear that half of the team is needed not for quality, but for the process.

Practical takeaway: 5 steps to implement an AI pipeline

If you are a marketer or CMO and want to repeat it, here is a short protocol. Not a “theoretical” one, but one that I took on three clients in 2024–2025.

1. Calculate the current production in hours and rubles. One standard release (video / post / landing page) - how much it costs, how long it takes. Non-estimated - upload timesheets and invoices for the quarter.
2. Subscribe to 3 tools: Veo (via Gemini Advanced or Vertex), ElevenLabs, Nano Banana Pro. Cumulative – $80–100/month. That's less than one hour for a good videographer.
3. Take one real brief of the week and do it in parallel: the team works as before, you work through the AI-pipeline. Compare the results blindly in a focus group of 5 people.
4. If the AI option wins or equals — transfer 30% of the volume to this pipeline. Not everything at once. Let the team see that this is not a threat - this is a redistribution of the load.
5. After 90 days recalculate unit-economics. I settled on ~−40% costs for content production and +200% speed. Your numbers will be your own, but the order will be the same.

What I still don’t know - and want to check

The most honest conclusion of this article is that I don’t know what will happen to the “strategic role” in 2 years. Now I am sure: the strategy, brief and final taste are up to the individual. After 2 years I'm not so sure. AI agents that read customer feedback themselves, form hypotheses and put them in a queue for testing are already working at the demo level. If they reach production-grade, my role will also shrink.

I don't take this as a threat. I perceive this as another pipeline that will need to be mastered before others. Just like I mastered Veo + ElevenLabs + Nano Banana before 90% of my colleagues.

If you have data on your AI-pipelines, write to Telegram @dipustovalov. I count in hours and rubles; other people’s data is also interesting.

Related materials: ROAS calculator, unit economics calculator, a complete guide to performance marketing.