Deconstruction of how to build an early-stage AI/ML startup

I fielded a question the other day in Johor, Malaysia about how much harder (or easier) it is to build an AI/ML startup. There are generally building blocks for any early stage startup, and they are:

  • team
  • product
  • invest initial capital
  • gain users
  • mentoring
  • follow-on capital

    Now I'll comment on the main differences in each category:

  • team: more technical or academic talent can help, but not necessary with open and available frameworks and MOOCs such as [this]( and [Andrew Ng's course](
  • MVP: the process of building an MVP is similar if not same. Timing could be just as similar.
  • invest initial capital: I initially thought that this could be slightly more, based on initial salaries; but at early stage there are no salaries to contend with. However, there are at least living costs.
  • gain users: same tactics and should be similar marketing costs as traditional startups. Some thoughts on tactics [here](
  • mentoring: more technical and AI specific needed
  • follow-on capital: definitely slightly more, based on salaries for hiring