Why BARK Makes AI Ads Deliberately Imperfect
At BRXND NYC 2026, BARK's Mikkel Holm walks through the work behind the company's deliberately imperfect ads. His team moved from complex generation workflows to treating Claude as a producer, asking for editable components and building a visual studio around the process. The aim is to give more people creative control while keeping human taste and ambition central.
Key takeaways
- Ask for the separate clips, graphics, voiceover and music so feedback can target the part that needs changing.
- Save the reusable campaign formula alongside the finished ad, making future variations easier to produce.
- Design the interface around the creative team's work and give the system real brand context.
- Use human taste to judge what works and ambition to push beyond the first acceptable output.
In this session
00:00 Why BARK makes deliberately imperfect ads
Holm introduces BARK's ad-making process and explains why ads that break its usual visual rules can still be useful creative experiments.
03:33 Claude as a producer with brand context
He describes treating Claude as a producer that coordinates tools with knowledge of BARK's humor, style and preferences.
06:07 Editable assets, creative control and reusable formulas
Separate assets and coded layouts let the team adjust individual components, make precise edits and reuse a campaign's structure.
07:47 Building a visual interface for the team
Holm explains why a chat interface is awkward for visual design and why the team is building a simpler studio around the same capabilities.
10:03 Inside BARK’s custom creative studio
A walkthrough shows brief inputs, creative hooks, visual editing and previews organized around the team's actual decisions.
11:29 Toy design, sales materials and product photography
Other examples include toy specifications, retail sales materials and combining real product photographs with AI-assisted image work.
12:37 The creative’s role: taste and ambition
Holm closes with the creative responsibilities he wants to protect: recognizing good work, knowing when to break rules and pursuing more ambitious ideas.
Recorded at BRXND NYC on November 5, 2026, at The TimesCenter in New York.
Written by
.png&w=3840&q=75)
Alephic Writing
Ideas for building the AI-native enterprise.
New essays on AI, marketing, software, and the organizations being rebuilt around them.
