OpenAPI Specification

But the data is all there on their website, so the entire machine-readable IKEA catalog is community reverse engineering.

There's a project called IKEA OpenAPI that documents four separate IKEA APIs.

a search API, a product catalog API, a sales item API, and a spare parts ordering API.

Somebody sat there and mapped out IKEA's internal endpoints from watching network traffic.

It's happened before.

Companies have adopted community standards because the community did the hard work of figuring out what the data model should be.

The IKEA OpenAPI project runs daily contract tests.

That's more diligence than some official APIs get.

A hot take that I have forming is that we're kind of looping back to the dawn of Web 2 where everything is kind of becoming just an API and we're sort of opening things, at least for agents, back up where it's pretty easy for an agent with curl even just to go get a lot of stuff done just on APIs.

Do you think we'll see more standards collapsing into kind of these behemoth standards like OpenAPI? Or will we still try to chip away at maybe the right agent-focused thing?

Yeah, so I think some new things do deserve to exist.

I think for the most part, the default is exactly what you said, where I think we'll just collapse on the same standards we've been using all the time.

OpenAPI, GraphQL, Nix, all these things are definitely no reason to invent a new thing for the most part.

I think this is really interesting and something I thought a lot about as just someone who's building tools, right? It's like what actually does, like what does something that actually does deserve to be built new look like, right? And I think what that looks like is when kind of the fundamental math you do of like how should this abstraction look and what should it optimize for, whenever that changes based on agents, like fundamentally what you want.

So like I think a really good example of this actually is just, the way we do pull requests, for example, the way we like get code merged, right? I think a lot of people will report, like if you start having like these like high velocity, many agents working at once, when it comes to this, I think you'll notice that like your agents just get stuck In rebase hell, they get just stuck having to...

So like for MCP, it's hard to justify, right? You're like, hey, they need a way to communicate, right? And it's like, okay, we have that for humans.

And so it's, it's been a very interesting time to see, you know, a lot of people, and I think people are starting to kinda, you know, realize that it, it, it's, it-- Just because Claude can go kinda prototype something, it, it doesn't necessarily mean that that's the right, you know, production ready system for them.

And, you know, really I, I, I like your approach of, like, why, you know, reinvent the wheel when you can find a solution that if, if there are a few things missing, you know, with that OpenAPI and with the MCP server and the ability to connect, like, you can add on, you can bolt on that missing functionality, you know, for your, for your own, you know, specific company's needs versus having to recreate the entire core and then get to those features anyways.

Like, it just, you know, I think a lot of times it does make sense when you can find companies that are very, you know, OpenAPI, you know, AI friendly.

You know, it gives you so much more power and, you know, uh, obviously, you know, you can have varying levels of cost.

I mean, you know, you got, you know, the ServiceTitans and the Procores and, you know, they have very, very expensive solutions or, you know, there's, there's more, you know, people like us who are much more affordable.

tiny go 0.42 we discussed this with someone else i think it was uh jamie tenna on the slack channel like why isn't tiny go versioned the same way as go versions right like why they have 127 why isn't it tiny go 127 you know what i mean

We talked about that with regard to OpenAPI, right? Because why is OpenAPI whatever not called 3.1? I wanted his OpenAPI

gen code to be like, oh, we're now at OpenAPI 3.1 or

whatever.

But I guess it gives you independence to do

We're testing private safety processing to strengthen safety monitoring.

For developers, Astra is available OpenAPI and is available in Amazon Bedrock.

OpenAPI standard pricing $10 million.

We talked about this.

Separate rates apply to cache reads and writes.

Exactly.

Before engineering starts, teams use schema modeling languages like Blueprint, Rammel, or OpenAPI, which many people still know by its original name, Swagger.

Oh, Swagger.

Yeah, I've heard of that.

The obvious objection first: we already have APIs.

REST, gRPC, OpenAPI specs, decades of tooling.

Why does an AI agent need a new protocol to call a database or a ticketing system? The answer is that traditional API integration was designed for developers, and agents are not developers.

Three gaps matter in practice.

In infrastructure terms, it decouples the lifecycle of your tools from the lifecycle of your agents, and decoupled lifecycles are how you avoid coordinated big bang deployments.

Semantics designed for models, not humans.

An OpenAPI spec tells a developer what an endpoint does.

A human reads the docs, handles pagination, interprets a four hundred twenty-nine, and writes retry logic.