Reuben

building at

[Reuben @ReubenKib]

@amirmxt this is exactly what I'm seeing as well.

but the gap on what to implement and the proper way to develop is only getting bigger as the models are getting better, ironic how that works

[Reuben @ReubenKib]

benchmarks are telling an interesting story. double the cost of Opus 4.8

[Claude @claudeai]

Introducing Claude Fable 5: a Mythos-class model that we’ve made safe for general use.

Its capabilities exceed those of any model we’ve ever made generally available.

[Reuben @ReubenKib]

@rosebeverly_ai at least 1

[Reuben @ReubenKib]

how many companies did apple kill in today's WWDC?

[Reuben @ReubenKib]

@tomsanee weekends should count. you need to restart.

[Reuben @ReubenKib]

@alokepillai @screenstudio as a paying screenstudio user, I find myself using cap more these days because of these issues

[Reuben @ReubenKib]

happy ap x swatch day

[Reuben @ReubenKib]

the city showed up last night

[Reuben @ReubenKib]

tokenmaxxing just got more expensive

[ClaudeDevs @ClaudeDevs]

Starting June 15, paid Claude plans can claim a dedicated monthly credit for programmatic usage.

The credit covers usage of:
- Claude Agent SDK
- claude -p
- Claude Code GitHub Actions
- Third-party apps built on the Agent SDK

[Reuben @ReubenKib]

This is just the beginning! Great to see more companies coming here

[Decagon @DecagonAI]

Decagon is opening an office in Toronto. 🇨🇦

Toronto has become one of the best places in the world to build AI, with an exceptional concentration of engineering talent.

We’re actively hiring across product, engineering, and GTM. ↓

[Reuben @ReubenKib]

regardless of how many tools I try, I always end up back in Cursor.

can't fully explain it. but the best tool is the one you keep reaching for.

[Reuben @ReubenKib]

there's been a lot of hype recently about HTML vs markdown for LLM prompts.

my findings earlier this year: when I need strict guidelines I use HTML. mainly for design and output structure. the model knows exactly what to render and where.

when I want freeform guidance, markdown. let the model interpret.

not everything needs to be one or the other.

[Reuben @ReubenKib]

A week ago I shipped a health AI agent that lives entirely in iMessage.

40 users in private beta. No app. Just text.

Yesterday Google announced their own health agent.

It's more important than ever to ship fast and pivot. I built and shipped mine in 10 days with Codex and Convex. Keep shipping and something will hit.

[Reuben @ReubenKib]

@hthieblot human in the loop is a myth

[Reuben @ReubenKib]

@masonwarner are you finding midjourney still better than gpt image 2 for frame reference?

[Reuben @ReubenKib]

GPT-5.5 Instant dropped. First thing I did was go through client projects and swap models where it made sense.

Customer-facing agents, automation flows with chained LLM calls, real-time copilots. Anywhere latency and cost compound fast.

That's the whole point of building modular agents. New model comes out, you update a config, done.

[Reuben @ReubenKib]

the n8n hype died but the tool didn't.

building a complex automation flow for a client right now. visual nodes, secure credential sharing, maintained third party integrations.

just because you could do it in code doesn't mean you should. nothing wrong with leveraging these tools if it makes sense for the project.

[Reuben @ReubenKib]

@forgebitz one of my favourite hobbies right now

[Reuben @ReubenKib]

@jaredpalmer @Xbox @asha_shar congrats. v0 was one of the first tools I used in the AI space and it changed how I build. excited to see what you do at Xbox.

[Reuben @ReubenKib]

@nutlope this is the way. been doing something similar with design patterns for AI builds.