Thanks for referencing the article. My read is that, the founders got started based on “whom they know” (i.e each other). Given the relationship and history that they had, they thought that they wanted to do something together (very effectual).
“Fast forward to 2017, when I was working at my second startup in the Bay Area, a satellite imaging company Planet. I reached out to Brian, who was still at Boeing. We both had a realization around this time that the next stage for our careers would involve working on interesting problems together,” Sharma says. “We made an agreement with each other that no matter what, we were going to quit our jobs and build a company.”
That problem turned out to be something that one of the founders came across just because he was in a startup that faced that challenge. (also seems effectual - almost accidental).
“There was a ton of interest at this time around how AI could be used in novel, groundbreaking ways like self-driving cars,” Sharma says “But one of the interesting things that was happening around me at Planet was the intention to build computer vision applications based on the large batches of data we would collect every day from 300-400 daily satellite images. The way our engineering teams would go about building these applications was by exploring what data they should even be using. There was this idea around labeling and it was a very new concept.”
This piqued his interest and he started to gather more information about the problem directly from the company he was working at. (a bit causal - he has a hypothesis that this is an interesting problem and this is validating it).
This realization was enough to spark the seed idea of a new business to pursue. Sharma started to use his days at Planet soaking up all of the information he could around this problem space, taking notes on how internally the team was exploring ways to go about building an ML infrastructure and scaling AI.
Then they kinda went out and, in a very lean startup manner, talked to a bunch of AI experts in the field who were dealing with this problem. These conversations helped them widen their perspective and solidify their understanding of the problem space. This is causal thinking, I think. Very deliberate efforts to validate their problem hypothesis.
So maybe it’s a mixture of both in this case.
It reminds me of Paul Graham’s article How to Get Startup Ideas where he said
Paul Buchheit says that people at the leading edge of a rapidly changing field “live in the future.” Combine that with Pirsig and you get:
Live in the future, then build what’s missing.
Labelbox seems to fit into this idea, the founders were living on the cutting edge, they noticed something that was missing (a good data labeling system) and built something around it.