A marketing campaign that once took at least three hours now takes just 10 minutes at one company after an employee built an AI assistant to do the work.
The company in question is Good Karma Brands, a sports media and marketing business with more than 550 employees working across audio, video, digital, and live events. Its portfolio includes ESPN-affiliated brands in major markets such as New York and Chicago.
For those local teams, developing business with advertising partners requires campaign materials tailored to the market and the potential partner. Before those materials go partner-facing, the sales manager and marketing team review the campaign deck.
In New York, assembling those decks was part of the job of Gina, one of the company’s marketing professionals. Ty Bauschek, Good Karma's Senior Director of Innovation and Sales Development, was helping lead the company's broader AI transformation when Gina showed him a system she had built to do the work differently.
She called it CAM, short for “Campaign Assistant Manager.” (The name is also a nod to New York Giants running back Cam Skattebo, one of her favorite players.) And the first live demonstration turned hours of manual work into about 10 minutes of effort.
Gina was not spending hours every day just staring at a blank presentation.
Good Karma's New York market already had the raw materials needed to build partner campaigns. The problem was assembling the right pieces for each opportunity.
Before a daily 4 p.m. review call, Gina would find and pull slides, bring the presentation together, edit the materials, search online for partner logos, and prepare everything for the sales manager and marketing team to review.
According to Bauschek, each campaign took two to three hours at minimum. At peak, the market might need three or four campaigns in a single day.
That made campaign assembly a serious capacity constraint. The deck was necessary, but every hour spent moving slides and logos was an hour unavailable for more creative or partner-facing work.
Or, as Bauschek puts it:
"Nobody's best at accumulating slides."
Gina knew the workflow well enough to see that its most time-consuming steps followed a recognizable pattern.
She first built the solution through a custom GPT, then developed it further using Codex. She indexed a vault containing the slides available to the New York market and created a way for CAM to assemble the relevant materials from a campaign input.
Instead of searching through the slide library and rebuilding the deck manually, a teammate could provide the file, add the partner logo, and have CAM pull the campaign together.
The system was useful because it reflected how Good Karma's New York team actually worked. It used the market's own presentation materials and followed a process built by the person who had been performing it every day.
When Gina first showed CAM to Bauschek, the result looked too easy.
"I thought it was a magic trick," Bauschek says. "I thought it was a rabbit-in-the-hat situation. Like, that can't be true. You're tricking me."
So Bauschek changed one of the campaign's investment calculators and asked CAM to try again.
The system handled the change.
Then came the before-and-after comparison. Bauschek says the campaign that CAM assembled took 10 minutes. Previously, it had taken three hours.
That was the moment the workflow became more than an interesting AI experiment. The new process had survived a real variation in the work, produced a usable campaign, and made the time difference impossible to ignore.
The New York system was built around one market's materials, so it was not a universal campaign generator Good Karma could switch on everywhere.
But the method could be adapted.
When turnover created a need in Good Karma's Chicago office, Gina traveled there and helped the local team build its own version of CAM using that market's campaign materials.
The expansion showed that the result was neither a one-time demo nor a tool tied permanently to one office. The underlying workflow could travel as long as someone took the time to understand the local inputs and adapt the system around them.
It also became part of Good Karma's larger AI transformation. The company was already rolling AI access and training out broadly across hundreds of employees. CAM gave leaders a tangible example of what could happen when an employee moved beyond using AI for isolated tasks and redesigned an entire workflow.
Good Karma’s story offers some useful transferable lessons in how to evaluate your own AI workflow opportunities:
This is what makes the case study useful beyond Good Karma Brands. Many companies have important work wrapped in hours of searching, assembling, formatting, and moving information from one place to another.
AI does not need to take over the valuable human decision at the center of that work to create enormous value. Sometimes the opportunity is automating everything people have to do before they can do the work they do best.
Did you enjoy this transformation story? Go deeper on how AI is reshaping work and business with The Artificial Intelligence Show. Each week, we break down what matters in AI and what leaders should do about it.