“Could you build a full commercial with no actors, no crew, and no cameras — just AI? We did. And it cost us exactly $855.07.”
What If…
At Cresta, we believe the future of go-to-market isn’t just AI-assisted. It’s AI-powered. So we asked ourselves: what if an ad could be created entirely by AI, from idea to final cut?
Instead of theorizing, we tested it.
The Process
Step 1: Personal Experimentation
It started with me personally testing some of the latest AI video tools like Runway and Veo-3.
One of my early tests? A bizarre clip of a sailor with a hook through his neck (you can see it here: LinkedIn post). It was funny, rough, and a little disturbing — but it proved the tools were powerful enough to do something bigger.
Step 2: Spotting the Trend
Soon after, I started seeing AI-generated brand ads go viral — from Kalshi, IKEA, and Liquid Death.
That’s when the lightbulb went off. If others could do it, why not us? And the reasoning came down to three things:
- Reach – these videos were breaking out, and we had a chance to get Cresta’s message in front of the world.
- Inspiration – we’re an AI-native startup, and experimenting with AI in marketing helps the entire team stay cutting-edge.
- Cost – compared to the two Super Bowl ads I’ve helped create (which were not cheap), this could be done for under a thousand dollars.
Step 3: Building the Team
From there, we pulled together a cross-functional group:
- Creative/Video team to own production.
- Content & Social team to shape the story.
- Me (marketing lead, instigator, and occasional guinea pig).
We kicked things off by watching the viral ads that inspired us. A CMO I worked for before, Brion O’Connor, always liked to start with “what good looks like” to ground the team in vision and expectations. I’ve taken that forward.
Step 4: Brainstorming Concepts
We then ran a wide-ranging brainstorm:
- Big Idea: Cresta helps customers gain a competitive advantage by radically transforming the customer experience.
- Creative Direction: Frame this transformation through analogies to past technology revolutions — moments when new tools gave early adopters an edge.
This is where the AI tools began bolstering our human creativity.
We used ChatGPT to generate a long list of analogies around “technological shifts that create competitive advantage.” Some worked well — others didn’t. (One fun but impractical example to actually generate video for: “sending an email vs. a carrier pigeon.”)
Step 5: Generating Video Clips
With analogies in hand, we began testing different tools: Kling, Runway, and Google Veo-3.
- Kling & Runway gave us interesting creative directions.
- But Veo-3 consistently delivered the best instruction-following and audio alignment — especially after we used ChatGPT to optimize prompts and follow best practices.
Since Veo-3 only allows 8-second clips, we built scenes piece by piece, refining the script in parallel with ChatGPT to ensure the narration was tight, clear, and cinematic.
Step 6: Voiceover & Character Design
Once the script was locked, we moved to voice:
- Used ElevenLabs’ latest text-to-speech to audition multiple voices.
- Settled on one narrator that matched the ad’s tone, then layered in two additional characters midway through to keep things dynamic.
Fun fact: In the moon scene, we leaned on Veo-3 as the core audio source, which made syncing lip movements to the AI-generated voices surprisingly crisp.
Step 7: Music, Editing & Assembly
For music, we first experimented with ElevenLabs’ new music generation model. But here’s the challenge: none of us are music specialists at Cresta. Getting the right feel, tempo, and mood out of AI music required more sophisticated prompting (just like the videos did) and proved much harder than we expected.
Ultimately, we pivoted and purchased licensed music from Artlist — which gave us exactly the sound we needed to tie the visuals and narration together.
From there, we stitched everything together in DaVinci Resolve.
This was the most iterative part of the process — testing, re-generating, and re-editing each individual scene until it hit the right tone. Every detail mattered: pacing, transitions, voice variation, and visual consistency.
Step 8: Launch
Once the ad was finished, we knew the work wasn’t done. A great video doesn’t go viral on its own — you need to be deliberate about distribution.
We expanded the team by bringing in our Head of Communications to lead media outreach and our Digital Marketing leader to manage paid amplification. Together, we coordinated a simple, easy, but effective launch plan.
Here were the major tactics:
- Multi-profile release on LinkedIn. We published the video across three main profiles, each with a distinct narrative:
Cresta’s company page → focused on the core message.
My profile (as CMO) → positioned it as a statement about the future of advertising.
Our CEO’s profile → highlighted it as an innovation by our team. - Cross-platform publishing. The video also went live on YouTube and X to reach broader audiences.
- Internal engagement drive. We sent out a calendar invite to employees encouraging them to like and comment on the launch posts, ensuring early momentum.
- Investor amplification. We asked our investors to help spread the word across LinkedIn and X, tapping into their networks for extended reach.
Paid promotion. We supported the organic push with ads on LinkedIn thought leadership placements, LinkedIn CTV, and YouTube ads targeted to our key audiences.
And the results started showing up quickly. In fact, someone already told us they saw the ad run on TV as they were walking down the street!
Working Rhythm and Timing
One thing worth noting: this wasn’t an all-consuming project. In fact, it was a side project layered on top of everyone’s day jobs.
- We met for just 30-minutes a week over 5 weeks.
- In between, various team members put in 4–8 hours each week pushing the work forward (things intensified as we got closer to launch)
Could we have finished in a few days? Absolutely. But treating it as a light, creative sprint gave the team space to experiment, fail, learn, and come back each week with sharper ideas (and get their other projects done).
And I think that iterative, low-pressure rhythm made the final product better.
The Budget Breakdown
Here’s exactly how the $855.07 was spent:
- Kling AI – $50 (ideation/testing, ultimately not used)
- Runway – $15 (ideation/testing, ultimately not used)
- Veo-3 – $430.07 (core video generation)
- ElevenLabs – $33 (narration & voiceover)
- Artlist – $350 (licensed music after AI trials)
Total: $855.07
Prompt Best Practices
One of the most difficult parts of this project was actually getting the prompts right.
AI video prompts aren’t simple — they can get long, detailed, and highly technical. It wasn’t just about describing an idea; it was about giving instructions like a director would on set.
Here are a few of the biggest learnings:
- Go beyond the idea. Don’t just say what you want to see — specify how it should be filmed. Include things like focal length, character positioning, camera angle, and motion.
- Tell the model what not to create. An exclusions list is just as important as the creative direction. It saves you from wasted generations.
- Expect limitations. Some things the models just won’t get right. For example: we wanted a team name written on a football field. No matter how many times we prompted it, the model refused to render it properly. In the end, we had to edit the team name in ourselves.
Here’s one of the prompts we used to kick things off:
In Summary, Why This Matters
- Speed → What normally takes months happened in weeks.
- Affordability → Budgets that once required six or seven figures can now fit inside four.
- Creativity at Scale → We tested more ideas in one afternoon than many agencies do in a quarter.
- Team Flow → Big things can happen even with tiny, consistent bursts of effort.
- Transparency → By showing the budget, we hope to inspire others to experiment.
The Big Question
This ad isn’t just marketing — it’s a signal. The tools are here. The workflows are real. The costs are accessible.
👉 The question isn’t if AI will change how we tell stories. It’s how fast.
Would you trust an AI-made ad to represent your brand?
In Closing
At Cresta, our mission has always been to help teams unlock their highest potential with AI. This ad is just one more step in that journey — showing that AI isn’t just behind the scenes… it’s ready for the spotlight.
BIG thank you to the team involved. Vladi, Steve, Alexa, Nicky, Devon, Annie, and Michael.
Read the original post on LinkedIn.
