Muse Wonder feels…great? Using it in @t3dotcodes with 10 subagents at once for code review, already found lots of issues left over by previous 6.1 Sol workers. And here's a voxel Colosseum made with Wonder xHigh. Pretty damn good.

I'm working on two apps. For members this week, I wrote about my setup based on @t3dotcodes and @poteto's amazing pstack system. Let me know if you want to see more of this!

I'm very happy with my hybrid @bot + @cursor_ai workflow for managing internal @macstoriesnet apps. Here's how it works, along with a few tips. It's an always-on (cloud-based), self-testing setup that I've been (mostly) driving from my phone for nearly two months. And it has made us way faster for articles and admin tasks. This is very much inspired by @poteto's excellent /pstack skills and tips (the interview with @mattpocockuk is a must-watch) and it's a series of techniques that fundamentally resonate with me and how my brain operates. I started 2026 with a goal: I just want to write and record podcasts; I don't want to be dragged down by inefficient tools anymore. So I created a series of web services for us, initially built with Codex, then redesigned by Fable in June. We've stress-tested these for months, including during WWDC and the Apple September event. Some notes: - I've created custom tools for our CDN, Apple Newsroom analysis and extraction, article search with @voyageai rerankers, Notion Markdown export and proofreading, and more. Millions of people have read articles that were written by @johnvoorhees and me, but managed via these tools. - There are outer and inner loops. The outer loop is @bot coordinating different bots and threads per-app. Each app gets its own thread, and the primary bot orchestrates across them. - The inner loop is a Cursor project. These are similar to Claude's new projects in that they're cloud based, there's a coordinator, and everything runs in a cloud VM. Inside a project, agents (with different models, like Opus for UI work) can work on different parts of our MacStories tools. - AFAIK only Cursor's projects support setting env secrets (Claude can only set API credentials in Projects, bummer). This allows agents to build and run the apps in their cloud sandbox, test them with real data, and it lets the QA bot come in and validate. - The parallelization of tasks is where @bot shines right now with multiple bots and threads weaving feedback in and out from a Cursor project. To the best of my knowledge, no other tool makes it this easy right now (I can approximate this with Hermes, but it's a lot of extra work and overhead). I just couldn't do all this with Dots either because it's just one bot. - I use pstack and verification skills religiously. I want my agents and bots to be constrained to check their work, operate independently, and escalate to me only when needed. - I can operate all this from…anywhere. I have added features to the apps during walks with voice mode (talk to primary bot, let it coordinate), and I've even kicked off sessions with webhooks from my Pebble ring: - The system watches itself with routines that scan logs and analytics for errors, hands off to Cursor cloud agents for review, which verify and open PRs This was a LOT of work to put together, but it works beautifully for us at this point! We're a team of two (plus our designer and dev), but I feel like our output has tripled this year (and the numbers prove it). I hope we get more options in this space in the future, but for now the Bot + Cursor combo is unmatched.

This was almost way too easy, @natfriedman @alexandr_wang 😮 I turned my Xteink eReader into an always-on Muse pocket display for Calliope (my Muse) to display itself and show status updates. And this attaches with MagSafe to my phone, so I can always see relevant updates from Muse by just turning my phone around. The setup flow in the Muse app was super smooth and you can just ask Muse to load or automate content on the display. I genuinely love the hacker-friendly aspect of this. Well done.

One of the parts of @bot that really works for me is the multitasking aspect of it all. My primary bot can spin up group threads with other specialized bots (coder, researcher, etc.), which can then spin up @cursor_ai cloud agents, and coordinate. Today, I deployed new features for one of our internal @macstoriesnet apps by doing voice mode with primary bot, which then orchestrated the rest. (Including a @NotionHQ board to keep track of everything across bots.) From a product perspective, none of the other hosted/cloud assistants (Muse, Dots, etc.) currently match this. Will be interesting to see if others follow this approach (and I think they should).

I gave all my different agents the same @NotionHQ board for all the projects I work on, and it's honestly been a lifesaver. (I'm also creating a LOT more things at once!) My story:

Seeing some WILD improvements to M5 Ultra performance for MLX after just one week. Measured recent oMLX upstream builds, token prefill has essentially doubled across the board with Flash-Next. Only one week has passed. Will need to update my review soon!

I reviewed the M5 Ultra Mac Studio with 256 GB of RAM. It’s the dream Mac for local AI agents. And I’ve gone local-only with Hermes. My story, with LOTS of interactive charts:

The latest oMLX release by @jundotkim is terrific. Finally can use a local model (qwen3.8-flash-next-q4) running on my M3 Ultra Mac Studio, from my iPhone (with Open Minis) to do "real" assistant work. Running at a respectable ~60tps. No cloud costs. Love it.

I've been working on this for 3 months, and here it is: My iOS and iPadOS 27 review: Customizable Liquid Glass, iPadOS, 15 system apps, and a big Siri AI chapter. And no, it's not a Snow Leopard release. Enjoy!

Finally got my @Pebble Index 01 ring and super excited to dive in! It looks great and I've heard great things about its AI integration with MCP as well as webhooks. Fun project ahead 😎 @ericmigi

Here it is: Our full interview interview with @ajambrosino about the Codex app. The history, design challenges, Computer Use, and more. If you love working in Codex, I think you'll enjoy this interview. Podcast link:

Another lovely update for @ChatGPT on iOS and Codex Remote: Customizable widgets! You can now add new widgets that let you change the buttons to anything you want for Chat, Work, and Remote – including Voice. Instant Home Screen material. I love this.

"The updated Polishing Cloth has dropped in price to $9 from $19. In a world of high-priced memory and storage, I’ll take the pricing wins where I can get them." @johnvoorhees 😂