Overview
At a senior-healthcare company, I planned an AI education program for adults in their 50s and 60s as a new business line built on our existing senior customer base. I ran the full cycle on my own: proposing the business, designing the curriculum, teaching, running retrospectives, and building the recruiting channel.
Problem
It was unclear whether an AI education market for people in their 50s and 60s even existed. In an external interview, a corporate executive in their late 50s flatly said, “There won’t be any demand,” and there was skepticism within the team about a high-priced class model.
Approach
Instead of deleting the negative feedback from the plan, I used it as the basis for changing direction and redefined the business as “a business for the buyer.” I designed small free classes, held at least once every 2 weeks, as a data-collection mechanism, and collected surveys from every cohort to feed into the next cohort’s curriculum and the results report. When paid ad recruiting for cohort 3 failed, I built and deployed an education brand website in a day as a new recruiting channel and a way to measure response (8 static HTML pages + 1 CSS file, hero video 1.8MB→468KB, fonts subset from 3.2MB→361KB, 0 touch targets under 44px).
Results
Proposed the business plan in April → ran cohort 1 (July), cohort 2 (August, 2 classes at an apartment complex, afternoon and evening), and cohort 3 (September): 3 rounds of classes, 20 participants in total, and 12 surveys from cohorts 1 and 2 (10 very satisfied, 2 satisfied, 0 dissatisfied). On the day it was sent, the cohort 2 results report was passed to the client (an apartment management office) as sales material, which opened discussions about a 2nd round of classes and tie-ups with neighboring complexes. 1 graduate moved into a job as an AI coach. Cohort 3 ads on Karrot, a Korean neighborhood app, failed with a 1% click-through rate and 0 sign-ups via the ads, so I replaced them with our own website as a channel. In October, 10 people applied for the first paid session, 9 of whom came through ads. At a local government workshop for resident autonomy committee members (about 220 people), I handled the booth and training materials, making it the first time a local government paid for our classes. I restructured the program into 3 tracks (AI for Everyone, Everyday AI, Smartphones) plus a CEO class.
Lessons Learned
When you keep counterevidence (a negative interview, skepticism within the team, a 1% click-through rate) and update your decisions instead of erasing it, that record becomes the plan for the next cohort. An education business is proven not by its curriculum but by repeated cohorts and collected surveys.