James Moschou

@landonepps Once the code works :) I haven't actually run it and I had to learn C++ as I went.

@jcmosc You should totally write a blog post! I suspect you know more about AttributeGraph than anyone else who can speak publicly about it. I guess it's a bit of a different field, but anything about the internals/architecture of AG would be super useful for Apple platform devs.

I recently discovered that I can use an LLM to automate Ghidra, and I asked it some questions about SwiftUI internals. The results far exceeded my expectations. This is the first task I've given an AI where it does a better job than I could.

@landonepps I'm not building anything to do with a user interface, but there are enormous parallels. Instead of SwiftUI -> DisplayList -> pixels, you can think of it as [Name TBD] -> Emit Instructions -> source code bytes

@landonepps I'm building a no-code app builder that will use an incremental computation engine, and I think the type system and expressivity of Swift is unmatched for what I want to be able to do.

For any Wordle fans out there, I created a similar daily puzzle... Find the number sequence in as few guesses as possible producthunt.com/posts/sequence…

I'm so excited to attend #apidays Australia next week as a speaker! My topic is "Mapping your business model to API design". Whether you're a tech person or from business, there will be something in it for you! Register here --> ticket.apidays.global/event/apidays-…

Date and time information is so common in APIs that they can make or break your API's developer experience - evaluate different approaches based on your use case from @getcriteria hubs.li/Q01ZxpXF0

Friend of mine @jcmosc has built a really cool tool for building flow diagrams that integrate with your OpenAPI spec 👀 Take a look at diagrams.criteria.sh

Try it out and let me know what you think! (And watch this space)

Because the focus is on API workflows, you can import an OpenAPI file and autogenerate flows much more rapidly than general-purpose tools allow. Since diagrams are intended to be published in public docs, they have a much more polished and professional appearance.

I've been a bit quiet on here, bootstrapping a collaborative design platform is definitely not easy... So I'm going to pivot away from collaboration and release some mini products where folks can get value in seconds, starting with... Criteria Diagrams diagrams.criteria.sh

🚀🚀🚀 Here's a quick roundup of what we released this May: 1. Sync Spectral rulesets 2. Configure API styles 3. Import Swagger 2.0 documents Keep reading on...

@PatThePM I've seen design teams conduct time-on-task studies, and work out the cost savings or efficiency improvements for the customer. Say a task used to take x minutes and now takes y minutes, and it's done 20 times a week. What does that add up to over the year for the customer?

If your company’s OKRs are primarily business outcome oriented, how do you get UX improvements prioritized that don’t move a business metric?

@Erwin_AI yeah fair. Though you could be like "hey when you get a minute I would love to chat about what you're working on" and let them get back to it

@jcmosc Haha I just don't wanna be "that guy" that randomly starts talking to someone in deep focus.

There is someone next to me in my coworking space that is using the chrome devtools to change Tailwind classes. The urge to tell him about Tailscan is UNREAL. Should I?

@Agustarson It definitely helps, but you have to be good at compartmentalising. You're job is to let the team arrive at a solution themselves, which means biting your tongue a lot.

Any engineers turned product managers out there? I'm curious to hear if it helps or hinders you (I know, it all depends). I manage API development and sometimes think I'd be in a better position if I had an engineering background.

Soon a large proportion of your API's consumers will not be developers, but AIs. We'll be writings docs, not for developers, but for language models. Maybe it's time to start using #aiexperience instead of #developerexperience?

3. AIs are excellent writing boilerplate code. This makes AIs great at writing API clients and "glue" code. Zapier says it takes about 2 weeks for a tool to integrate into their platform. An AI-powered Zapier could onboard any tool immediately.

Been thinking about the future role of AI automation in APIs. I think we're gonna see AI write code on the consumer side first. The provider side of API development is still very much led by humans.