The Plan Failed Before the Camera Did
- A rescue is not proof a plan was sound, it is proof the rescue worked.
- A shoot that depends on one person and one day is a bet, not a plan.
- Redundancy is cheap compared to a shoot that nearly returns nothing: a weather day, a backup shooter, a reviewed shot list.
- AI belongs in a production plan as one more line with its own review gate, not as the backstop you count on when a plan breaks.
Business of Home ran a piece last week by Fred Nicolaus called How AI-generated imagery is quietly reshaping the design industry. It tells the story of Chase Kerlin, founder of the furniture company Olive & Wool, who planned a shoot in the Colorado mountains. He hired a photographer and shipped his new line hundreds of miles to get there. Then the photographer missed the flight, and the day was overcast, which hid the mountains. The piece puts the cost at roughly $35,000, a trip that came close to producing zero photos. Kerlin took some basic shots himself. A friend of his, Michael Bashour, ran them through generative AI tools and made them usable. That rescue became a company, HookPic. Most people will read this as AI beating photography. I read it as a production planning story, because that is the job I did for years before I built and led a team.
What actually went wrong on that Colorado shoot?
Two things went wrong, and either one could sink a shoot. Chase Kerlin's photographer missed the flight, and the day in the Colorado mountains was so overcast you could not see the mountains. Kerlin had rented an Airbnb, hired the photographer, and paid for freight, and Business of Home puts the trip's cost at roughly $35,000. It came very near to producing zero photos.
I do not know what Kerlin had planned for before he left, and the article does not say. What I am reading is the story as told: a shoot where one missed flight and one bad weather day stood between a finished shoot and nothing. In about seventeen years building and leading a creative team, that is the exact shape of failure you learn to build around. Flights get missed. Weather does not care about a call sheet. A plan that cannot survive either one is not a plan, it is a hope with a budget attached.
Is this really a story about AI beating photography?
No. AI did not outperform a photographer on this shoot, it caught a shoot that had already gone wrong. The trip came very near to producing zero photos, and the founder's own basic shots became usable after a friend ran them through generative tools. The win belongs to the recovery, not to some contest between two methods of making a picture.
That distinction matters because of what people will take from it. A good save at the end of a bad day proves somebody found a way out after the plan broke, nothing more. I have watched teams mistake the rescue for the lesson before, more than once. The rescue gets the praise. Nobody goes back and asks why the rescue was needed in the first place.
What should a production team take from this?
A plan that only works if nothing breaks is a bet, not a plan. The fix does not require new technology, it requires the same discipline that has always separated a shoot that survives a bad day from one that does not. Build in redundancy before anyone flies anywhere.
- A weather day on the calendar, not just the one date that looked good when travel was booked.
- A second shooter or a clear plan B, so one missed flight does not end the production.
- A shot list the whole team has reviewed before anyone leaves for the location.
- A review step once images come back, so a problem gets caught on day one instead of discovered at the edit.
None of that is exotic. It is the ordinary cost of insuring a shoot, and it is cheaper than a $35,000 trip that nearly returns nothing.
Where does AI belong in a production plan?
AI belongs in the plan as one more line with its own review gate, not as the backstop you reach for when the real plan collapses. Four Hands' VP of creative services put it plainly: AI is not consistent and exact enough yet, and asking it to change something in an image it generated can turn into an endless loop of redirecting, redirecting, redirecting.
Treat it the way you would treat any vendor: reviewed, not trusted blind. A marketer quoted in the piece calls this the number one AI trap, thinking you can go from zero to one hundred in one step. His fix is the same discipline that belongs on any shoot: legal looks at it, executives look at it, creatives look at it, product marketers weigh in. AI can rescue a bad day. It cannot replace the planning that keeps you from needing a rescue, and the team that forgets that difference will keep mistaking a good save for a good plan.
Frequently asked
What happened on the Colorado furniture shoot that AI ended up saving?
A furniture founder planned a shoot in the Colorado mountains, hired a photographer, and shipped his new line hundreds of miles, on a trip that cost roughly $35,000. The photographer missed the flight and the day was overcast. The founder took basic shots himself, and a friend ran them through generative AI tools and made them usable. That rescue became an AI imagery company.
Does this story prove AI can replace a photographer?
No. The story is not AI beating photography, it is AI catching a shoot that had already gone wrong. The trip came very near to producing zero photos, and the founder's own basic shots saved it after a friend made them usable with generative tools. The lesson is about planning for a bad day, not about one image-making method outperforming another.
How should a production team put AI in a shoot plan?
Treat AI as one more line in the production plan with its own review gate, not as a substitute for planning or a rescue you count on. Four Hands' VP of creative services has said AI is still not consistent and exact enough, and one marketer warns against trying to go from zero to one hundred in a single step. Legal, executives, creatives, and product marketers still need to look at the content before it ships.