| >You get to discover that for yourself. We spent seven years delivering ML products to enterprise, and are building something based on what we discovered. However, this is not a photo sharing application where the use case is very narrow and one developer is enough to pin-point the need and execute on it. This is a meta-product that enables people to build products. There is some complexity in that the ML landscape is varied and it is precisely why many are trying to do that. We can discover part of this and we did, but we want to uncover more by having users try it out. A concrete example: during the first months of the pandemic, our colleague had thirty students with machine learning final year projects. These were in a hot COVID spot. We had total lockdown, and these students did not have the possibility to go to university to access computers. They also did not have means to buy a costly workstation just for the project. We onboarded them on the platform as "early users" so they could train machine learning models, deploy them, and build applications that used them, and as a result, were able to graduate on time. Now, I don't know what you think of that approach, but preventing 30 students from losing an academic year is not a bad thing. Did the platform suck in some ways? Absolutely. They graduated on time, though. 30 * one academic year = 30 years saved because that product, even in its primitive form, existed. We can either develop in a void and wait until the product is complete and perfect to show to the world and allow everyone to use it, but I don't think anybody who actually shipped useful products in their life ever does that. Whenever someone asks how they could focus more when building their product, even a personal project, I tell them to get the right kind of users. Early users help uncover the unknown, and help focus on some of the knowns. For example, you have a long backlog. Without users, prioritizing can be done but is tricky. With users: after the 20th user complains about a bug/UX, you know what your next issue/ticket is going to be. The result is faster development on what matters, and the consequence is instead of shipping the product to the wider audience in a year, it'll be in 6 months. Having early users helps tear down things we might have ignored and accelerates bringing this value to more users. |
Could those 30 students have gotten GCP access and carried out their assignments? Probably. I wouldn't portray myself as a savior while trying to sell a product. I'd say that those 30 students were doing you a favor. (In the same way Autodesk, Microsoft etc are giving away student licenses to get you into their eco-system. It's really not a new thing)
But, to my original quote that you seem to have disliked - you are a small company (I saw 10-ish people?) that's entering a highly populated and competitive market with the assumption that your early adopters "need your product badly". I can only assume you have one helluva product. I'd say in 2-5 years you'll have the answer (not that it has to be bad news).