AI for Designers:
5 Practical Scenarios with FLUX, Sora, and GPT-5

I'm Mikhail, a 32-year-old designer. I've freelanced for eight years. Half my projects are graphic design — branding, packaging, key visuals for ads. The other half is motion — short social videos, intros, pre-visualizations. I started seriously integrating AI into my workflow in spring 2025. By early 2026, I had a solid tech stack, and it became clear: AI doesn't replace designers. It changes their speed and raises their ceiling.

Below are five scenarios I use weekly on actual commercial projects. We'll cover specific tasks, which models I use for what, and how much time and credits they cost. Plus, a separate section on what AI still can't do well. That's important; I'm not selling you some fantasy about 'vibe-coding design.'

Scenario 1. Mood Boards with FLUX, Not Pinterest

Building a classic mood board used to mean two hours on Pinterest and Behance, manually picking 30 references, narrowing down to 8-10 finalists, then compiling them in Figma. That's an hour of billable time for each client, or ₽2000-3000 in wasted budget just to 'start understanding what we're doing.'

Now, I open Quantium and generate 8-12 mood boards in 10 minutes using FLUX 2 Pro. Here's the prompt formula:

This isn't the final visual. It's a working visual hypothesis for client discussions. Before, I'd bring 6 stock images and try to explain, 'This direction, but with these colors.' Now, I bring 12 unique shots tailored to their specific brand. Clients agree on a direction faster because they see the result, they don't have to interpret it.

Savings: 2 hours → 15 minutes per project. For 5 brands a month, that's 9 extra hours in my pocket, or ₽18,000 in additional margin.

Learn more about prompt engineering in FLUX in our guide to 12 photorealism techniques.

Scenario 2. Campaign Pre-Viz with Sora 2

Pre-viz for ads used to mean a Figma storyboard: 8 frames, descriptions, arrows. The client would nod, we'd go shoot, and then find out the 'mood wasn't right.' The cost of a mistake on set? ₽200,000-500,000.

Now, I do pre-viz with Sora 2. For 15-20 credits, I get a 5-10 second video with my scenes. It's not final production, but the aesthetic, camera movement, colors, and overall vibe? All there. The client sees 'how it will be,' not 'how it could be.'

A real-world example: a pilot series of creatives for a tea brand. Here's the prompt for Sora:

The client saw it, said "that's exactly what we need," and approved the production budget. Without AI, this stage would take two weeks of approvals. With Sora, it's two days. Often, the final AI video is so good we use it as social media creative without reshooting. Learn more about Sora 2 in our Sora vs. Veo comparison.

Scenario 3. Logo Variations: Gemini Image + FLUX

When a client asks for "10 logo variations," I used to make 4 concepts and refine each for two hours. Now, the process is different: 3 concepts by hand (it's still my craft) plus 7 AI variations on each to broaden the search.

For logos, I use Gemini Image and FLUX 2 Pro. Gemini understands references better. Show it three logo examples in the right aesthetic, and it consistently delivers similar options. FLUX handles form and textures better without references.

One key thing: AI isn't great at drawing letters. If the logo has text, we still build the final version in Illustrator. AI helps with the pictogram, mark shape, color combos, and overall metaphor. At the presentation stage, clients see 10 directions, not 4. That changes how professional they see us.

Find detailed logo case studies in our FLUX logo tutorial.

Scenario 4. Scripts via GPT-5

I'm a motion designer, not a scriptwriter. When a client sends "we need a 60-second video about our service," I used to get stuck on structure: where to start, how to develop it, where to put the CTA. Now, I completely delegate that block to ChatGPT-5.

Here's the prompt that works:

GPT-5 gives me three options. I pick the best, tweak the tone to match the brand voice, and go into production. One script takes 25 minutes instead of two hours. It's not "AI replaced the scriptwriter"; it's "I've got a scriptwriter in my pocket for 8 credits."

Scenario 5. Voiceovers for Mockup Videos via TTS

When you show a client a video draft without a voiceover, perception drops 50% instantly. Voiceover is half the emotion. I used to either record myself on an iPhone or hire a voice actor for 1500-2000₽ per minute.

Now, it's ElevenLabs TTS in Quantium. I pick a voice (for Russian content, I grab ones marked "native"), paste the script text, and get audio in the right tone in 30 seconds.

Pro tip: For a business tone, I add "(calm, confident tone, medium speed)" in parentheses at the start of the text. For emotional videos, it's "(warm, friendly voice)." This changes the render more than you'd think.

When clients first see a video with finished voiceover, the impact's way higher. Often, after a presentation like that, they approve the final version without any changes. Learn more about TTS in our article on AI podcasts.

What AI Can't Replace Yet

Anyone who says AI will replace designers is either selling courses or isn't a designer. Here's what I still do by hand — and will for at least another 3-5 years:

• Typography. AI doesn't get clean letterforms or kerning. Any text-based logo, any lettering — that's Illustrator, by hand.
• Brand Identity & Guidelines. A 60-page brand book, tokens, usage rules — that's brain work, not a model's.
• UX Research. User interviews, JTBD, scenarios — that's about people. AI helps analyze transcripts, but it doesn't replace the researcher.
• Negotiations & Presentations. Defending a solution to a client? Always the designer.
• Final Layout. Pixel-perfect work in Figma, exporting for formats, handing off to production.

AI speeds up option generation 5-10x and takes care of routine visual searching. After that, the designer's the curator, director, and decision-maker.

Cost / Value: What It Costs & What It Gives You

Here's my credit spend for one average commercial project (brand visuals with a video, 3 weeks' work):

At 3000₽/hour, I save 57 000₽ in project time. My credit usage easily fits the Pro plan. If you do 5-6 projects a month, it pays for itself tenfold on the very first one.

The main thing isn't client savings. It's that as a freelancer, I can take on 2-3 more projects a month without sacrificing quality. That's what really changes income, not just "hours saved."

How Clients React to AI in Design

The most common question I get: "Do clients know you use AI?" They do. Here are a few observations from a year of using it.

Corporate clients are pragmatic about AI. They care about results, deadlines, and budget. If I win on those two metrics, there are no questions. Sometimes they even explicitly ask, "Can you show us AI options during the concept phase?" It's becoming the norm.

Small businesses and startups love saving money. I can take on projects from them I'd previously turn down because of budget. AI makes my rates affordable for clients who'd otherwise pay a freelancer with far less experience.

Design agencies I contract for are the most demanding. They need AI steps to not show through in the final product. So, I do 100% of the final layout by hand; AI only lives in mood boards, previews, and drafts.

Here's the key: don't show clients raw AI outputs. The final artifact should always be polished by hand. AI is a process, not a product.

What's Changing in the Profession

Over the last year, I've noticed three trends reshaping the design profession:

• The barrier to entry for visuals is dropping. Any non-technical founder can now make a decent mood board with FLUX. This means a designer's value will shift to curation, strategy, and final refinement.
• The value of taste is growing. AI generates average results, and to stand out, you need to tell good from mediocre. This isn't about technical skill; it's about a sharp eye and deep understanding.
• Speed is the new quality. Clients who once waited 5 days for a concept are now surprised if it takes more than a day. That sets new expectations, and you simply won't keep up without AI.

The profession is changing. It's not disappearing; it's evolving. And designers who master AI now will be in a much better position by 2027-2028 than those who keep working 'the old way'.

Related resources: 12 FLUX Prompt Techniques, Sora vs Veo Comparison, Midjourney Alternatives, Work Gallery.

Frequently Asked Questions

Read also

Gemini Image for product photography

12 min

Negative prompts: how to remove artifacts

13 min

Sora 2 vs Veo 3: What to Pick

10 min