Michael Doise

iOS dev | Built Perspective Intelligence (on-device AI), Beyond The Gallery & more at @TechopolisApps | Tech, sci-fi, gaming | Making iOS accessible to everyone

[Michael Doise @mikedoise]

Hello everyone. I am very excired to announce AFM Studio, the best way to test AI models and providers on macOS and iOS. The project is open source and contributions are welcome and encouraged. You can learn more and try it out by visiting the repo at https://t.co/4zoYkMwN9k We have support for the system model, PCC, and Core AI models currently. This even includes our own model registry that will be updated often. Enjoy! #buildInPublic, #OpenSource, #AppleAI

[Michael Doise @mikedoise]

@The_CozyDev I do read them especially these days. I like to see what new features people add especially if I'm looking for better iPhone Duo features.

[Michael Doise @mikedoise]

@BlindGuyTech It looks like a cool device, but it looks like the Siri experience is not great on it. Big shocker I know, but it looks like a good speaker and a cool screen. I think it will be good in the future, but it has some rough edges. @JoannaStern review showcases this stuff.

[Michael Doise @mikedoise]

Does the HomeView or even the Apple Watch for that matter really need an App Store if Siri and app intents can make app experiences from your iPhone? For example, you could open an Audible app intent to show the books in your library and play them, or use the same technique as CarPlay. Download once and use anywhere.

[Michael Doise @mikedoise]

Perspective Intelligence 3.0 is amazing. I'm working on the router some and then we should be good to go. I had to re-work the LiteRT implementation to do better with memory management and tool calling, but I think we are good to go there.

[Michael Doise @mikedoise]

I think that LLMs are more amazing than anything we've ever invented. They can do so much so quickly and we really don't take advantage of their power. It is time to focus on what's possible with AI and leave conventional software behind.

[Pratik Gandhi @pratikg]

Local AI Builders Friday # 4

The list is now 122! A lot of you nominated people under # 3, thank you ❤️

I added 11 more amazing builders this week:

1) @dhh is an OG! He is working to transform Omarchy into a Local AI distro. Starting with Omarchy 4.6 - it’ll try to select the optimal AI model for your hardware and lets you share AI models across your tailnet (ie use your DGX Spark to provide AI capabilities to your netbook)

2) @0xSero (@exolabs) got DeepSeek running entirely offline on his pack-of-portable-compute (that he literally carries around in a backpack!) 🤯 and is getting 200+ tok/s

3) @ItsmeAjayKV is doing his own Local AI R&D (kernels, quants, etc.) using just a 3090 and 64GB of RAM - and is hands-down one of the best open weight model news aggregators/accounts to follow

4) @christopherowen runs TP4 on 4 DGX Sparks cabled in a ring (no switch!) and implemented his own one-shot RoCE all-reduce algorithm getting 13-19 µs allreduce latencies (vs 75-86 µs for NCCL). Check out his repo with 279 experiments (he also logs his misses, with numbers!)

5) @tpae is building @OsaurusAI, free, open source (MIT) software for running local AI on your devices. Check out his Apple @ Work appearance this week (his pitch: “own your AI. At work too.”)

6) @dealignai (also working on Osaurus AI!) makes the case that since real tasks are input-heavy, using prompt processing on 4 DGX Sparks instead of a 512GB M5 Ultra makes sense (similar $)

7) @tayarndt and 8) @mikedoise are building Perspective Intelligence at Techopolis - an on-device AI chat assistant for Mac, iPad and iPhone. @tayarndt is also behind DictateMute - free, open source software that automatically mutes you on Discord when you’re dictating to Claude Code

9) @thelogicalgate uses his 128GB M5 Max MBP and 128GB RTX Spark to run local LLMs and is working on his own inference engine for RTX chips (to run on Windows on Arm)

10) @aijoey is running some of the frontier open models on just 2 DGX Sparks and documenting everything (configs, benchmarks, even building tools like Spark Doctor to help others get the most out of their Sparks)

11) @rickmanelius has 3 Macs + 2 DGX Sparks all networked together using Tailscale so he always has access to 3 or 4 local AI models (and can decide which one to use for different tasks)

Follow them and follow the full list here: https://t.co/DLJgziZNPM

Who should be in # 5?

[Michael Doise @mikedoise]

There is one small feature that would make iOS and iPadOS be the best operating systems on mobile. They should allow for multiple apps to play audio and use the microphone. I should be able to be on Discord and mute there and talk to my ChatGPT Dot while also watching YouTube. and maybe play a video game. All of that should just work like it does on a Mac.

[Michael Doise @mikedoise]

I didn't want a HomeView... and then I watched the reviews, and now I kind of want a HomeView.

[Michael Doise @mikedoise]

@thsottiaux I’d love a live activity in the iOS app to see the status of working codex sessions.

[Steven Scott @BlindGuyTech]

Google’s Gemini glasses will ship on the 10th of December and will work on iPhones - but there’s a catch. Read the latest now: https://t.co/m9TJFRcq5F

[Michael Doise @mikedoise]

@ijustine @forthewlvs I just can’t believe someone would even say something like this. Your content is amazing! I always look forward to your reviews and videos.

[ChatGPT @ChatGPT]

Your dot now has…

⚡️ Faster replies
💬 Less jargon
🫡 Better follow-through

And, they can now follow your custom instructions. Tell your dot in the thread things like how to behave or which actions need your go-ahead.

“always talk in lowercase.”
“Show me a draft before sending any emails.”

You can also set custom instructions and rules in Customize > Behavior on mobile, or through your dot’s menu on web and desktop.

[Michael Doise @mikedoise]

@AnxiousHolly I’m not the biggest Techwoven fan. The one for my 17 Pro Max broke after 6 months.

[Michael Doise @mikedoise]

Google DeepMind needs to now release Gemini 4 Oxygen so we can breathe after all of these new model releases.

[Michael Doise @mikedoise]

I think we also take longer documents and split them in chunks. We basically use compaction to summarize the document. You can then use different filters to re-examine the document to get specific details based on the limits of the context window. Asking for a summary of key points in a document would be a completely different model call compared to summarizing about action items.

[Michael Doise @mikedoise]

I've noticed that I really don't hit guard rails much anymore since 26.4 when chatting in my app, or in Locally AI with it. It used to be all the time. The other nice thing about the Foundation Models library though is that you can use adapters for MLX, Core AI, and other providers to bring in models like Qwen or Gemma 4. That gives a lot more context and power with the same framework. he advantage of Apple's model is that it is fast and doesn't require long warm ups.

[Michael Doise @mikedoise]

@Dimillian I keep mine folded.

[Michael Doise @mikedoise]

@xburak @RevenueCat makes paywalls so easy to make. Getting users is the hard part.

[Grok Bot @bot]

Introducing Grok Bot 1.0. AI teammates that work alongside your whole team.

Share Bots, collaborate with them in Slack and Teams, invite them to meetings, and watch them learn and become more helpful.