WORK ABOUT FUN CONNECT
AI-first Hackathon
Creative Direction Nate Eames
Video Editor Raffy Coletta
Context
During our company offsite in July 2025, our team that spanned across folks from finance, marketing design, engineering, and sales all came together around a singular goal — develop a sick brand anthem video with the use of AI in a single day, with $100 and six hours on the clock.
Problem
This wasn't just a time and budget constraint. It was a creative one. At the time, AI video generation was genuinely unpredictable — it didn't understand physics, struggled with continuity, and had a telltale quality that made AI-generated content immediately identifiable. Most of the team had never prompted a video model before, let alone edited across multiple tools like Runway, Midjourney, Sora, and Veo.
Our creative director had sketched a rough storyline, but it was clear we needed to be flexible to work within what AI could actually do.
Video Editor Raffy Coletta
Context
During our company offsite in July 2025, our team that spanned across folks from finance, marketing design, engineering, and sales all came together around a singular goal — develop a sick brand anthem video with the use of AI in a single day, with $100 and six hours on the clock.
Problem
This wasn't just a time and budget constraint. It was a creative one. At the time, AI video generation was genuinely unpredictable — it didn't understand physics, struggled with continuity, and had a telltale quality that made AI-generated content immediately identifiable. Most of the team had never prompted a video model before, let alone edited across multiple tools like Runway, Midjourney, Sora, and Veo.
Our creative director had sketched a rough storyline, but it was clear we needed to be flexible to work within what AI could actually do.
Exploration
We split the team into parallel tracks — three hours, six people, each testing different scenarios and platforms to see what held up. What we learned quickly: the original concept of a worn-down healthcare worker wasn't just an overplayed narrative, AI's limitations would work against the story.
That realization became the pivot. Instead of fighting the slop, we leaned into it.
We split the team into parallel tracks — three hours, six people, each testing different scenarios and platforms to see what held up. What we learned quickly: the original concept of a worn-down healthcare worker wasn't just an overplayed narrative, AI's limitations would work against the story.
That realization became the pivot. Instead of fighting the slop, we leaned into it.
Process
The first 2 - 3 hours were about exploring and prompting. Collecting half-decent clips, sharing prompts along the way to maintain visual consistency.
The next 3 hours, I worked on piecing everything together, while others went to find the perfect background music, and record and re-record the voice over to be just right.
The first 2 - 3 hours were about exploring and prompting. Collecting half-decent clips, sharing prompts along the way to maintain visual consistency.
The next 3 hours, I worked on piecing everything together, while others went to find the perfect background music, and record and re-record the voice over to be just right.
Solution
We reframed the narrative around levity — letting the absurdity and glitchiness of AI-generated visuals become part of the creative voice rather than something to hide. The ridiculousness wasn't a bug; it was the tone. The final film embraced what the tools were actually good at, turning the constraints of the medium into a distinct aesthetic point of view and delivering a brand anthem that felt surprising, self-aware, and alive.
We reframed the narrative around levity — letting the absurdity and glitchiness of AI-generated visuals become part of the creative voice rather than something to hide. The ridiculousness wasn't a bug; it was the tone. The final film embraced what the tools were actually good at, turning the constraints of the medium into a distinct aesthetic point of view and delivering a brand anthem that felt surprising, self-aware, and alive.
We won Best Wow Factor at the offsite — but the real validation came after. The film was A/B tested against the current homepage video, and within two weeks of going live, across 30k+ visitors, the AI-first version drove a 12% increase in engagement and interactions. What started as a scrappy hackathon bet ended up outperforming production.
The intangible results was a genuine confidence in AI as a creative tool. Work that would have taken weeks to concept, produce, and approve was live and in-market within a month. That speed doesn't just change timelines — it expands the possibilities.