Regarding the first part - I'm quite used to hearing that question. But usually it's not Google or similar, rather a big player in your market than can, actually, decide to build your product. Take Solid for example - it's very likely Snowflake and/or Databricks will build (or buy) a competing product. So, it's a legitimate question. The issue I have with that question is that my answer is obvious - they already know the answer when they ask it.
So why ask it?
Secondly, the fact that the foundational models are expanding into everyone's territory is a really interesting dynamic. Been watching it for the past two years as new models keep coming out. The conclusion - build a product that gets much better as the models get better. Don't fix gaps in those models because the gaps will likely close.
On the first part, what is your answer? That you'll just do it better than them? (Though, if that is the answer, I don't actually hate that question, though it's a little bit of a trick. Because I'd want a startup to approach it that way - say you're just going to build a better thing; you're going to line up with them, toe to toe, and beat them; and don't give some bank shot answer that distracts you from just building a good product.)
And yeah, you periodically see VCs say similar stuff, about how people should be building for the models that exist in five years, not the ones that exist today. Easier said than done though, of course.
1. One, we'll do it better. Not because we have better people, but because our people are 100% focused on solving this really well.
2. Most companies have multiple databases / warehouses, and big vendors are notoriously bad at supporting their competitors' tech. So we're Switzerland here and we get to keep all the gold.
The first answer always makes a lot of sense to me; the second one always feels kinda overly optimistic. They're really good at supporting *their* tech, and so the tradeoff of "good general multi-lingual tool" vs "very good one language tool" has always felt like a tough one.
What about the absolutely dismal track record of actual attempts by Google specifically? Did Google Drive squeeze out Dropbox? Did Google Meet kill Zoom? Did Stadia steamroll Steam? (Out of innate sense of mercy, I'm not going to ask about Google Plus, Allo, or Cardboard.)
Yeah, and that's definitely another argument that you could make, that Google is a beatable competitor even if they are trying. (That said, I do think there are some interesting structural dynamics to this stuff too. Drive and Meet surely have taken a big chunk from Dropbox and Zoom, but the markets there are so big that everyone can still be successful. And on stuff like Plus, I've always sort of assumed that one of the reasons those products don't work is because they try to make very general versions of it. They aren't opinionated in any way, because they need to serve a billion people. So in both cases, it's still kind of a big market thing.)
I was listening to this post with Safari’s text-to-speech feature while putting dishes away, but it kept pronouncing “chatbot” wrong which seems so hilariously emblematic of Apple’s failure to keep up with AI.
Yeah, I’ve talked about this before, but I think the “how big is your market?” question is one of the most misunderstood ones in tech. On one hand, it matters way more than people think, because very little matters more than having a lot of people you can sell a thing to (and you can’t overcome the inverse, of having nobody to sell to). On the other hand, it matters way less than people seem to think, because new stuff often doesn’t seem to map onto existing markets. So the question really should be, “how are you drawing the market, and how big is it with that new drawing?”
Regarding the first part - I'm quite used to hearing that question. But usually it's not Google or similar, rather a big player in your market than can, actually, decide to build your product. Take Solid for example - it's very likely Snowflake and/or Databricks will build (or buy) a competing product. So, it's a legitimate question. The issue I have with that question is that my answer is obvious - they already know the answer when they ask it.
So why ask it?
Secondly, the fact that the foundational models are expanding into everyone's territory is a really interesting dynamic. Been watching it for the past two years as new models keep coming out. The conclusion - build a product that gets much better as the models get better. Don't fix gaps in those models because the gaps will likely close.
On the first part, what is your answer? That you'll just do it better than them? (Though, if that is the answer, I don't actually hate that question, though it's a little bit of a trick. Because I'd want a startup to approach it that way - say you're just going to build a better thing; you're going to line up with them, toe to toe, and beat them; and don't give some bank shot answer that distracts you from just building a good product.)
And yeah, you periodically see VCs say similar stuff, about how people should be building for the models that exist in five years, not the ones that exist today. Easier said than done though, of course.
Our answer is multi faceted actually:
1. One, we'll do it better. Not because we have better people, but because our people are 100% focused on solving this really well.
2. Most companies have multiple databases / warehouses, and big vendors are notoriously bad at supporting their competitors' tech. So we're Switzerland here and we get to keep all the gold.
The first answer always makes a lot of sense to me; the second one always feels kinda overly optimistic. They're really good at supporting *their* tech, and so the tradeoff of "good general multi-lingual tool" vs "very good one language tool" has always felt like a tough one.
What about the absolutely dismal track record of actual attempts by Google specifically? Did Google Drive squeeze out Dropbox? Did Google Meet kill Zoom? Did Stadia steamroll Steam? (Out of innate sense of mercy, I'm not going to ask about Google Plus, Allo, or Cardboard.)
Yeah, and that's definitely another argument that you could make, that Google is a beatable competitor even if they are trying. (That said, I do think there are some interesting structural dynamics to this stuff too. Drive and Meet surely have taken a big chunk from Dropbox and Zoom, but the markets there are so big that everyone can still be successful. And on stuff like Plus, I've always sort of assumed that one of the reasons those products don't work is because they try to make very general versions of it. They aren't opinionated in any way, because they need to serve a billion people. So in both cases, it's still kind of a big market thing.)
I was listening to this post with Safari’s text-to-speech feature while putting dishes away, but it kept pronouncing “chatbot” wrong which seems so hilariously emblematic of Apple’s failure to keep up with AI.
Yeah, I’ve talked about this before, but I think the “how big is your market?” question is one of the most misunderstood ones in tech. On one hand, it matters way more than people think, because very little matters more than having a lot of people you can sell a thing to (and you can’t overcome the inverse, of having nobody to sell to). On the other hand, it matters way less than people seem to think, because new stuff often doesn’t seem to map onto existing markets. So the question really should be, “how are you drawing the market, and how big is it with that new drawing?”
Yeah, this was something we ran into a lot with Mode. Your market is whoever you're competing with, whether you like it or not.