BLUF (Bottom Line Up Front)

  • Cloud and open source technologies have lowered the up-start cost of tech ventures.
  • But the success of (most, if not all,) businesses still hinges upon a real value proposition.
  • The dual and the genesis of a real value proposition is a great pain point.


The Lucky PhD Co-Founders


Montse Medina and her co-founders would have fitted squarely in the category of your stereotype of PhD students. Give them a hardcore rocket science or any other technical challenges, they would happily chip away at it. But a business problem? That is not what they had in mind originally.


So how did they build their nifty AI startup, Jetlore, and sell it to PayPal so quickly?  As Montse described, without (i) the convenience of cloud, and (ii) the fortune of hitting a real pain point (one that hurts so much that customers are willing to pay and share data with them to have it solved), they would not have gotten where they are today [1].


Back-End Is Cheaper. Algorithm Is Free. So What…


In his widely popular article published back in 2011, Why Software Is Eating the World [2], Marc Andreessen had given us a hint, the back-end infrastructure is being commoditized. The on-demand, pay-by-usage model would be here to stay. The implication of this lowering of cost is we will see more cloud-based software firms being stood up.


Seven years later, what we are seeing is software is still eating the word [3], albeit on an even bigger scale, enabled by the pervasiveness and availability of cloud services provided across the entire solution stack, from SaaS, PaaS, to IaaS (e.g. Microsoft is a case in point).


Now with AI being associated with Industrial Revolution 4.0, it is only logical to cook up more companies in the cloud, and sprinkle with some form of AI here and there (another reason one has gotta love AI, it is so broad, all-encompassing, and full of poorly defined, jargon-y terminologies, it would not be a stretch to label yourself as an AI company as long as you are utilizing some technique that remotely fits within the realm of AI).


The best part about that? Many building-block-level AI technologies (by technologies, I mean things like different machine learning, deep learning algorithms and so on) exist in the form of open-source implementation. Therefore, the cost of adding AI ingredients to a cloud startup can be fairly small with the right technical expertise (e.g., you could go out and fetch some open source libraries, bake them into your platform, sell your repackaged goodie to customers that live on another layer of abstraction; this is a perfectly fine monetization model.)


Whether that product is something the end user wants to consume though, is a completely different story. And this is when the other important factor of success comes in.


Did You Find A Problem? A Big Problem?


Without a real pain point, the best technology is only a lab project.


Without customers actively looking for solutions, the funding of your shiny new cloud, AI startup will eventually dry up.


Montse was lucky that Jetlore stumbled upon a burning problem. But the real lessons learned there is, for all technically inclined (future) founders out there, start with a business problem.


(More to come in a future follow-up post)


Full disclosure: The author works for a vendor of automated machine learning platform; I have no intention to pick on cloud or AI. It just happens I was inspired by Montse Medina’s talk on Jetlore as well as Dean Patron’s talk on Microsoft cloud. Thanks for reading.




[1] Montse Medina, presentation on Stanford campus, 2018





Users who have LIKED this post:

  • avatar


  1. This is an absolutely great share of your insights, and I really enjoy it. As more and more algorithms and library are open-sourced, the cost of learning and applying has been lowered a lot, which indeed empowered many technology companies to be less technical-centered, fast-growing and profitable. Although I do truly appreciate those working hard in the frontier of science, I think for some people they just want to apply those less-advanced but quite applicable technologies into the real business world and start to make a larger, visible influence. And I think you and me are one of those!

    1. Thanks for your feedback, xwang01.
      Having been a consultant in my previous life, I am used to thinking we can map every problem to a 2-by-2 matrix… In this case, we have (i) on the one axis, difficulty level (in terms of technical feasibility and implementation challenge): Easy –> Difficult; and (ii) on the other, value/impact: low ROI –> high ROI. The art is to identify those problems in the Easy & High ROI quadrant and convince the prospect/client that is what they should absolutely focus on first and foremost.


Comments are closed.