AI bootcamps for college students
A multi-year program pairing college students with real companies on eight-week applied AI projects. Over a thousand learners came through it.
What it was
Built and ran in partnership with HubSpot Academy. We sourced students from universities across the country and ran them through eight weeks in two halves. Closer to a paid internship than a class: students were paid through the program, so the companies they worked with took on no cost and no risk.
How the eight weeks worked
The first four weeks were intensive training, several hours a week, with deliverables, homework, and resources to work through every week. We held students accountable with tracking, submissions, and attendance. Anyone who missed a session watched the recording.
The back four weeks paired each student with a real HubSpot customer who’d signed up with an actual need. Students built them an AI-enabled marketing campaign: tools set up and usable, an agreed volume of content, and a roadmap for what came next.
The campaigns used the full HubSpot toolset. Blogs, marketing emails, automations, forms, CTAs, all built with AI-enabled tooling. Clients arrived with something specific in mind, usually an upcoming event, a product launch, or general nurture marketing, and students ran the build for things like the lead-up to a conference, a virtual event, or an executive dinner.
Part of the deliverable was teaching the client to use the tools themselves, so the work didn’t evaporate when the student left.
Where the companies came from
I worked with HubSpot Academy, the customer success team, and other departments with direct customer contact, pulling data on good-fit accounts. Industry didn’t matter to us. What we wanted were smaller teams who owned marketing tools but weren’t activated on them.
The pitch worked in both directions. Companies got real project work at no cost, and the program got those accounts using tools they already paid for. The interest was never the hard part.
What changed from the first cohort to the last
The first one was messy. Around 300 students and a little over 200 companies, and we were figuring out student management, scheduling, and how to actually deliver payment at the end while the thing was already running. The content was unpolished because I’d never run it before. HubSpot’s AI tools were early too, several in beta, so things broke constantly and we were stressing the system in ways it hadn’t been built for. Everyone was learning live.
By the end it was a different program. We had a real process for building the trainings, informed by what worked and what to drop from previous runs. The biggest single fix was scoping the capstone projects down. The first ones were too ambitious, and the overwhelm showed up as imposter syndrome in students who were doing fine.
Later cohorts held steady around 200 to 300 students and 150 to 250 companies each. We added an advanced track for graduates of the basic one, and by then we had students and clients coming back for another round.
What I learned
Watching a student go from unsure to capable over eight weeks is the most satisfying work I’ve done. A number of them got part-time and full-time offers from the very clients they’d been assigned to, which was the clearest signal that the work was real and not an exercise.