AI-Nation bootcamp
A 12-week online bootcamp that teaches middle and senior developers to build AI-driven development into their work. Three cohorts launched, 110 seats sold.
Key figures
Money
- Revenue3 cohorts, 110 seats: $32K, $33K, $39K
- ~$104K
- ROMIby cohort: 7.5x, 5.1x, 6.5x
- 5.1–7.5x
- Average order valueby cohort: ~$915, ~$1,120, ~$850
- $850–1,120
- CACcohorts 1 and 2: +80% while reach grew 50% and revenue stayed flat
- $121–219
- Upgrade from single modules to the full coursecohorts 1 and 2
- 7/8 · 21/29
- Refundscohort 1
- 2 of 35
Learning
- Bootcamp CSI, average across cohortsaverage of cohorts 1 and 2: 87%, 89%
- 88%
- Completioncohort 1 → cohort 2
- 60% → 71%
- Practices applied at work12 in cohort 1, 14 in cohort 2, 26 across both
- 26
- Average webinar attendancecohort 1: 21 of 33 watched more than half of the lessons; cohort 2: 16 of 29
- 22
Idea
Developers already use AI assistants, but few have fully built AI into their work. The bootcamp closes that gap with practice: each lesson is built around a real case, and the theory is only as deep as the case requires.
The audience is middle and senior developers and their teams.
The programme runs for 3 months: 15 lessons as 2–3-hour webinars, one or two a week. The content came from the experts: an ex-CTO of T-Bank and director of the SymbioWay recruitment centre, an ex-PM of Yandex Cloud and a Google staff engineer; I designed the bootcamp. I ran the product as product manager and LXD, shared the producer role with the CEO and set tasks for a team of four (SMM, video editing, production, project management). The product launched with no audience of its own, through cold placements in Telegram channels and partner integrations.
Product decisions
Marketing
Checked demand for topics against developers' chat histories: ran them through Claude and ranked the topics by how often they came up.
Reach grew from ~200K impressions across 5 channels in cohort 1 (35 seats) to ~300K across 15 channels in cohort 2 (29 seats). YouTube thumbnail CTR went from 2–3% to 6–10% after a competitor analysis.
The funnel was fully tagged from ad impression to deal, with analytics connected; deals were recorded in amoCRM under their source tags.
Learning
Designed the programme with ADDIE and backward design: positioning, tone, structure and 15 lessons built around practice, with just enough theory for the case; the cases were the core entity the programme was built from.
Calibrated the audience level against an available target-audience base and a recruiting agency's CV base. 16 senior engineers, team leads and C-level managers validated the programme iteratively.
For the LMS I built a library: detailed lesson descriptions, every file from each recording (a cleaned-up transcript, the slides, the videos), practice briefs and extra materials.
Improvement cycle
The programme changed after each cohort on feedback from learners and experts; bootcamp CSI grew from 87% to 89% between cohorts.
After cohort 1: added agent harness and security topics; the experts added more worked examples.
After cohort 2: added MCP development for business and managing agent systems (Pi, Orca).
In the LMS library: clickable timestamps, detailed lesson descriptions and a mobile-friendly layout.
Student feedback
- Average bootcamp CSI across cohorts: 88%
- Team leads and CPO/CTO-level buyers rated the programme highly and enrolled their teams
- A team lead from cohort 2 brought about 15 people to cohort 3
- 3 graduates of cohort 1 enrolled again in cohort 3
- A cohort 1 student was promoted and credited the bootcamp techniques