Startup Studio: AI-Accelerated Building & Validation

I designed and teach this course on AI for Startups at Columbia University. Student teams build real products with AI tools, and learn how to build products that people love.

What goes on in the course

Students form teams and build a startup across the semester. Teams launch real products and acquire real users, and they learn along the way how to build products that people want, need, and love. The course rests on three pillars.

  • Agentic engineering Using AI tools to rapidly build software that is reliable and secure.
  • UX techniques Methods for advancing product-market fit, including user interviews and usability testing.
  • Startup knowledge Growth hacking, presentation skills, fundraising, and how founders operate.

How a semester runs

Each team carries one idea through the semester. The work moves through five stages, and every stage produces something a person outside the class can look at.

  1. 1

    Form the team and the concept

    Teams come together and settle on the problem they want to work on.

  2. 2

    Validate the pain point and people's desire for a solution

    Wizard of Oz tests, vaporware sites, paid ad campaigns and other demand experiments, measured rather than argued about.

  3. 3

    Build the MVP with AI tools

    A working product, built fast with AI coding and design assistants.

  4. 4

    Interview users and iterate

    UX research feeds the next version. Iteration history is part of what gets graded.

  5. 5

    Demo Day

    Each team presents to peers, instructors, invited alumni, hiring representatives, and potential mentors.

Demo Day

Each team gets fifteen minutes to present and answer questions. The pitch covers the problem, the solution, validation results, iteration history, and where the product could go next. A panel of industry guests gives feedback, and there is usually a networking session afterward.

Materials

The full syllabus lives on GitHub, and it covers the reading list, the grading breakdown, and what students need to supply themselves. Read the course syllabus.

Slides go up after each class, so the class materials repository fills in through the semester.

Who's teaching

I am Ken St. Clair, Instructor at Columbia University, and I created and teach this course. Outside the classroom I work with teams to provide trainings/workshops, help them get codebases AI-ready, and develop AI workflows. Check out what I offer on the home page.

This headshot was AI generated from multiple real at-home headshots

Book a call

30 minutes, video or phone

Loading the calendar

Or email me at ken@stclair.ai