“Practice what you preach”. Reminded of Sam’s Startup playbook here, specifically the section on Product. Full playbook - playbook.samaltman.com
“A great product is the only way to grow long-term. Eventually your company will get so big that all growth hacks stop working and you have to grow by people wanting to use your product. This is the most important thing to understand about super-successful companies. There is no other way. Think about all of the really successful technology companies—they all do this.
You want to build a “product improvement engine” in your company. You should talk to your users and watch them use your product, figure out what parts are sub-par, and then make your product better. Then do it again. This cycle should be the number one focus of the company, and it should drive everything else. If you improve your product 5% every week, it will really compound.
The faster the repeat rate of this cycle, the better the company usually turns out. “
Well summarised Dharmesh. Connecting this and your previous blog, clearly folks who have traversed the hard path of chunks/embeddings/memory management see this as a simple but specific implementation of functionality. For others who haven't seen the nuts and bolts will make this approach more sticky till frameworks like LLM start covering them with abstraction. Still a very early path for settling down with approaches. Interesting as it forces us to keep pivoting on ease/cost and time to build.
"Net result: The switching costs go up the more people use these features as a result of which the moat around OpenAI widens. That’s the strategic benefit."
Really liked the perspective on the defensibilities going up. It's almost like company owners and execs willfully ignore any benefits when its too "painful" to switch/rebuild. Do you think they will raise prices once competitors get shaken out or just keep adding new features and charging for those?
Can't speak for them, but my sense is that Sam and team are solving for the long, long-term. I don't think they're deliberately making moves to drive out competition so that they can then command higher prices. I think they're just building the best A.I. platform they can build.
I saw your post about a Substack API. I built SubstackAPI.com. Let me know if you’d like to use the endpoint or need help building an integration for your Hubspot CRM. Would love to help :)
Great post Dharmesh, you talked about how openAI is creating a moat for themselves.
Would you write an article on how someone who is just starting out can build a moat in their product as everyone in community has access to same GPT models, APIs etc What's the best strategy one should follow ?
“Practice what you preach”. Reminded of Sam’s Startup playbook here, specifically the section on Product. Full playbook - playbook.samaltman.com
“A great product is the only way to grow long-term. Eventually your company will get so big that all growth hacks stop working and you have to grow by people wanting to use your product. This is the most important thing to understand about super-successful companies. There is no other way. Think about all of the really successful technology companies—they all do this.
You want to build a “product improvement engine” in your company. You should talk to your users and watch them use your product, figure out what parts are sub-par, and then make your product better. Then do it again. This cycle should be the number one focus of the company, and it should drive everything else. If you improve your product 5% every week, it will really compound.
The faster the repeat rate of this cycle, the better the company usually turns out. “
Thanks for the comment. He's totally right.
Dharmesh - Great insight, and definitely something for startups to think about incorporating into their growth strategy!
Well summarised Dharmesh. Connecting this and your previous blog, clearly folks who have traversed the hard path of chunks/embeddings/memory management see this as a simple but specific implementation of functionality. For others who haven't seen the nuts and bolts will make this approach more sticky till frameworks like LLM start covering them with abstraction. Still a very early path for settling down with approaches. Interesting as it forces us to keep pivoting on ease/cost and time to build.
"Net result: The switching costs go up the more people use these features as a result of which the moat around OpenAI widens. That’s the strategic benefit."
Really liked the perspective on the defensibilities going up. It's almost like company owners and execs willfully ignore any benefits when its too "painful" to switch/rebuild. Do you think they will raise prices once competitors get shaken out or just keep adding new features and charging for those?
Can't speak for them, but my sense is that Sam and team are solving for the long, long-term. I don't think they're deliberately making moves to drive out competition so that they can then command higher prices. I think they're just building the best A.I. platform they can build.
Thank you I appreciate the reply. Excited for your next post. Thank you for doing this.
I saw your post about a Substack API. I built SubstackAPI.com. Let me know if you’d like to use the endpoint or need help building an integration for your Hubspot CRM. Would love to help :)
Great post Dharmesh, you talked about how openAI is creating a moat for themselves.
Would you write an article on how someone who is just starting out can build a moat in their product as everyone in community has access to same GPT models, APIs etc What's the best strategy one should follow ?
Am I the only one that doesn't actually see that image? It made me crazy curious :)
There was an issue when I first posted. The image should be there now.