A filmmaker of the lonely AI, a foreign generative video model and a national television place during one of the biggest sporting events of the year. It is not the intrigue of a science fiction film; This is the new reality of advertising, and it was created in just 3 days.
TLDR:
- First of the genre: A Commercial generated by AI For the event platform Kalshi was created using the Google generative video model, Veo3and distributed nationally during the NBA final.
- Reduction of drastic costs: The filmmaker AI PJ Accetturo created the announcement of 15 clips in just 3 days, resulting in a reduction in costs estimated at 95% compared to traditional commercial production. It costs $ 2,000
- The workflow “Gemini-to-Veo3”: The announcement was produced using a simple but powerful process in 4 steps: Rugueux script, prompt generation supplied by AI with Gemini, generation of videos with Veo3 and final edition in standard software like Adobe Premiere.
- The future is agile: The project signals a passage to smaller and highly qualified creative teams taking advantage of the AI to produce quickly and to high volume content and adjacent to a high volume.
- Human competence is Always Key: Despite the technological leap, the creator stresses that the professional taste, the experience of director and, above all, the writing of comedy are the new moats of the creatives in the era of the AI … and I agree!
In a historic moment for advertising and AI, an advertisement generated almost entirely by AI was broadcast on national television this week during the NBA final. The announcement, for the Kalshi event market, was an original idea of ”Filmmaker” Ai Accetturo, which was hired to produce a place on people who place betting on everything, from sports to current events.
The announcement is here.
The result is a testimony to the rapid progression capacity of generative video technology. In a detailed article on X, Accetturo has unveiled the effective shocking process behind the advertisement, which exploits the powerful new Google video text model, Veo3. This achievement, coming only a few weeks after the public beginnings of VEO3, underlines the frightened speed at which AI is adopted for commercial use with high issues.
Creative work flow powered by AI: script on screen in days
The Acpetturo process, which, according to him, has generated more than 30 million views on various projects in just three weeks, is a masterclass in human-Ai collaboration. He breaks down his viral video workflow into four simple stages:
- Write a rough script: The process begins with a fundamental creative idea.
- Use Gemini for prompts: Google Gemini are used to expand the script in a detailed shooting list and generate specific prompts for the video model.
- Generate with Veo3: The prompts are introduced in Veo3 (via Google Flow) to generate raw video clips.
- Edit in Capcut / Premiere: The final clips generated by AI are assembled, timed and polite using standard video mounting software.
Co-writing with a machine: development of vision
For the Kalshi spot, the creative process began by establishing some key dialogue extracts. From there, Accetturo collaborated with Gemini to invent what he called “10 wild characters in disarticulated situations” to deliver the lines.
“I co-write with Gemini,” explains Accetturo, “asking for ideas, choose the best and transforming them into a simple script.” This partnership allows rapid ideas, mixing a human creative direction with the limitless imagination of a Great language model.
The invitation is the new staging
The critical translation layer between the script and the screen is the prompt. Accetturo has refined a specific method for this, using Gemini to convert each photo of the script into a very detailed paragraph for Veo.
“I then ask Gemini to take the script and convert each plan into a detailed VEO prompt,” he notes. “I always tell him to return 5 invites at the same time – more than that and the quality begins to slip.”
The key, he underlines, is to treat each invite as an autonomous instruction, offering Veo the full context each time to maintain consistency in character, framework and tone.
Here is an example of a detailed prompt used for the ad:
A medium -level shooting, filmed like Raw Street images on a crowded band from Miami at night. An old white man at the end of the 1960s was paid with confidence on the sidewalk, surrounded by tourists and clubs. He smiles from one ear to the other, his belly which comes out proudly from a cropped pink t-shirt. He wears extremely short neon green shorts, white tube socks, beaten sneakers and a cowboy hat in solid foam with glitter on it … While walking, he turns slightly towards the camera, still at mid-Essai and shouts with all confidence and joy: “ Indiana obtained this dog in ‘Em!' ‘.
Tips, tips and net profit
Working with Veo3 requires some initiate tips. ACCETTURO advises the execution of prompts in “fast mode” and iteration quickly. If a generation is not perfect, it suggests sticking the original prompt in Gemini, asking for specific changes and trying to try. He also notes some current quirks, such as occasional unwanted subtitles and the need to use sentences like “screaming at the top of their lungs” or the All Caps dialogue to arouse the cries of the characters of the AI.
Although VEO3 does not yet support coherent characters on several planes, the descriptions very detailed in each prompt can create a strong illusion of continuity.
The most amazing statistics of this experience is efficiency. “It took around 300 to 400 generations to obtain 15 usable clips,” writes Accetturo. “A person, 2-3 days. It is a reduction in costs by 95% compared to traditional advertisements. ”
The future of advertising and “creative ditch”
Accetturo quickly stresses that this revolution does not mean the end of creative professionals. “Just because it doesn't mean that someone can do it,” he said. “I have been a director (since) over 15 years. Brands always pay a premium for taste. ”
He envisages a future dominated by small agile teams producing “viral and adjacent content each week”, reaching 80 to 90% of the impact of a fraction of the cost.
So what is the defendable competence for filmmakers and advertisers in this new paradigm? According to Accetturo, these are not technical prowess but creative instinct. I agree and it is a good proxy for a human + AI future.
“The most precious competence of entertainment and advertising is the writing of comedy”, “ He concludes. “If you can make people laugh, they will look at the full ad, get involved with her and some of them will become customers.”
The announcement of the Kalshi NBA finals is more than a simple intelligent advertising; It is a dispatch from the future of media creation – a future that arrives quickly.
The newsletter of “filmmaker Ai” pj accetturo is there: https://pjace.beehiiv.com
Jean-Marc is a successful AI activity manager. It leads and accelerates the growth of the solutions fueled by AI and launched a computer vision company in 2006. He is a recognized speaker during IA conferences and a MBA of Stanford.
