Today I stumbled upon a team that has built a client and a few service providers using NIP-90: Data Vending Machines.
They detailed two milestones on the project's page that are a perfect example of the Before and After pictures that I can't help but point them out here.
The project is aiming at creating an AI system that generates images and logos.
Before
First take: using L402, the project is hitting an endpoint that requires payment to complete. This means, the client is integrating with a specific endpoint, paying and receiving a result.
After
At the end of the first milestone, they mention they are interested in exploring what this would look like using NIP-90: Data Vending Machines.
What happens?
Instead of integrating specifically with an endpoint, which could change it's price, lower it's quality, go out of business or simply stop being the best possible provider for these type of jobs.
Instead of this client tightly-coupling with a specific endpoint, they use NIP-90 to communicate with an infinite market of service providers who try to get the job done for them.
No endpoint need apply.
Guess what? PhotoBolt completely migrated over to DVM!
X vs Y axes / Free-market vs centrally-planned designs
If it's not abundantly obvious by now, the first approach demands from the service provider a complete output that is acceptable, and hopefully, the best in the market.
Thus, the business behind the endpoint needs to vertically integrate, which means, they need to try and be the best at each step required to deliver a complete result.
If it's not the best in the market, the client will have to, at some point, shop around until they find a new endpoint to integrate with to buy this type of jobs from.
Rinse and repeat.
The new alternative that DVMs unlock is that there is no need to acquire services from a vertically-integrated service provider, which allows services providers to specialize in being the best possible at something widely specific and let Nostr and Lightning to be the glue that ties everything together.
Today I stumbled upon a team that has built a client and a few service providers using NIP-90: Data Vending Machines.
They detailed two milestones on the project's page that are a perfect example of the Before and After pictures that I can't help but point them out here.
The project is aiming at creating an AI system that generates images and logos.
Before
First take: using L402, the project is hitting an endpoint that requires payment to complete. This means, the client is integrating with a specific endpoint, paying and receiving a result.
After
At the end of the first milestone, they mention they are interested in exploring what this would look like using NIP-90: Data Vending Machines.
What happens?
Instead of integrating specifically with an endpoint, which could change it's price, lower it's quality, go out of business or simply stop being the best possible provider for these type of jobs.
Instead of this client tightly-coupling with a specific endpoint, they use NIP-90 to communicate with an infinite market of service providers who try to get the job done for them.
No endpoint need apply.
Guess what? PhotoBolt completely migrated over to DVM!
X vs Y axes / Free-market vs centrally-planned designs
If it's not abundantly obvious by now, the first approach demands from the service provider a complete output that is acceptable, and hopefully, the best in the market.
Thus, the business behind the endpoint needs to vertically integrate, which means, they need to try and be the best at each step required to deliver a complete result.
If it's not the best in the market, the client will have to, at some point, shop around until they find a new endpoint to integrate with to buy this type of jobs from.
Rinse and repeat.
The new alternative that DVMs unlock is that there is no need to acquire services from a vertically-integrated service provider, which allows services providers to specialize in being the best possible at something widely specific and let Nostr and Lightning to be the glue that ties everything together.
PS. Here are milestone 1 and milestone 2