Case studyAI TrainingWorkforce education company, university AI programs
How I taught Upright's live AI automation course and got asked back for the next cohort.
7 weeks
of live classes, one hour every week
Live builds
from a blank screen, with Q&A
Cohort 2
asked back after the debrief
01The situation
Upright is a workforce education company that runs tech and AI courses with universities, Rutgers among them. Its AI automation course pairs self-paced modules with one live hour a week.
Most learners had already used ChatGPT. What they couldn't do yet was get it to run a task without them sitting there. The live hour had to take a mixed-skill class from a typed prompt to a working automation, and still handle whatever they asked.
02What we built
- I taught the live hour every week for seven weeks. Upright wrote the curriculum and the slide decks. I ran the demos and the Q&A and kept the class on pace.
- When learners said a finished build taught them less than watching one come together, I switched to building from a blank screen, live. Sessions covered AI extraction in Zapier, prompts with structured outputs, AI agents, a lead-management agent and scraping LinkedIn with Apify.
- When someone asked about Claude Code or scraping, I showed it on screen, then brought the class back to the week's material. Between classes I answered questions on Discord.
03The results
7 weeks
of live classes in the first cohort, starting February 2026
Asked back
to teach cohort 2 after the debrief with Upright's program lead
1 proposal
an advanced course for learners who wanted to go past the beginner material
Details come from the course onboarding, the class recordings and the post-cohort debrief. Upright wrote the curriculum. I taught it.
People learn AI fastest when they watch a working automation get built and then build it themselves the same day. My team workshops run the same way, on the team's own work.