No surprise that AWS’s focus this year was mainly on Generative AI.
Having worked there I would say that was not the focus at the beginning of the year, and although everything that they presented brings a lot of value and is really cool, some stuff seemed rushed. Even the org that I worked in, completely scrambled to present something that had Generative AI in it and shifted their focused completely from what we were building 12 months ago (FOMO when looking at OpenAI and their recent success).
Anyways, out of the things that did not seem rushed, I noticed Code Whisperer’s new capabilities specifically ones revolving around processing code bases and answering questions about them.
Y Combinator funded a company that literally does the same thing and although I think that’s a great idea, I just don’t see how startups like this one can survive after AWS releases the same thing.
What I’m trying to get at is that it seems to me like bigger enterprises have a very very big advantage since the name of the game for generative AI is data, and lots of it, optimized over time.
Not saying that startups shouldn’t do Generative AI, or that they won’t be successful in doing so, but I’m a bit skeptical when I see all these bets being placed.
Just food for thought, obviously these fellas have a lot more experience than I do in investments.
Also everyone is talking about Open AI because they have in house LLMs developed and trained over time. No one seems to be talking about AWS, Google Cloud & Co - Amazon is literally collaborating with a handful of companies and we all now have access to like 5-6 different LLMs through Bedrock which I assume will be an approach that other companies will take as well.
So another thing to consider for startup founders in my opinion is - why don’t we build another AWS Lambda today? The answer is pretty clear, there are some products out there that are similar from some startups but none of them are billion dollar companies. I think that the same thing will happen shortly with all of this hype, where it will be extremely easy for an enterprise to release these products way faster than a startup would given the amounts of available data.
Speed was one of our main advantages remember? Well not so much in this case anymore.
Use Generative AI to solve real problems but keep in mind that if you just base your entire vision and business on that, you might fail fast since you’re never going to have the same amount and quality of data available to you to customize and improve an existent model (I’m assuming not all of us are PHDs that would be able to create a revolutionary model from scratch).
A lot of these Gen AI companies look like features that AWS can slap out of their existing services in weeks (and have done just that), because AI has become some much more accessible given the right amount of data. They actually probably obliterated 3-6 startups that got funded from the past 1-2 batches just at this reinvent alone…
P.S - you’re gonna tell me that they invest in founders not ideas. That’s cool, I hope those founders who were more successful than I was at getting into YC will consider these points while pivoting to new ideas…
Having worked there I would say that was not the focus at the beginning of the year, and although everything that they presented brings a lot of value and is really cool, some stuff seemed rushed. Even the org that I worked in, completely scrambled to present something that had Generative AI in it and shifted their focused completely from what we were building 12 months ago (FOMO when looking at OpenAI and their recent success).
Anyways, out of the things that did not seem rushed, I noticed Code Whisperer’s new capabilities specifically ones revolving around processing code bases and answering questions about them.
Y Combinator funded a company that literally does the same thing and although I think that’s a great idea, I just don’t see how startups like this one can survive after AWS releases the same thing.
What I’m trying to get at is that it seems to me like bigger enterprises have a very very big advantage since the name of the game for generative AI is data, and lots of it, optimized over time.
Not saying that startups shouldn’t do Generative AI, or that they won’t be successful in doing so, but I’m a bit skeptical when I see all these bets being placed.
Just food for thought, obviously these fellas have a lot more experience than I do in investments.
Also everyone is talking about Open AI because they have in house LLMs developed and trained over time. No one seems to be talking about AWS, Google Cloud & Co - Amazon is literally collaborating with a handful of companies and we all now have access to like 5-6 different LLMs through Bedrock which I assume will be an approach that other companies will take as well.
So another thing to consider for startup founders in my opinion is - why don’t we build another AWS Lambda today? The answer is pretty clear, there are some products out there that are similar from some startups but none of them are billion dollar companies. I think that the same thing will happen shortly with all of this hype, where it will be extremely easy for an enterprise to release these products way faster than a startup would given the amounts of available data.
Speed was one of our main advantages remember? Well not so much in this case anymore.
Use Generative AI to solve real problems but keep in mind that if you just base your entire vision and business on that, you might fail fast since you’re never going to have the same amount and quality of data available to you to customize and improve an existent model (I’m assuming not all of us are PHDs that would be able to create a revolutionary model from scratch).
A lot of these Gen AI companies look like features that AWS can slap out of their existing services in weeks (and have done just that), because AI has become some much more accessible given the right amount of data. They actually probably obliterated 3-6 startups that got funded from the past 1-2 batches just at this reinvent alone…
P.S - you’re gonna tell me that they invest in founders not ideas. That’s cool, I hope those founders who were more successful than I was at getting into YC will consider these points while pivoting to new ideas…