Founder, @DaywardAI. I help companies position and explain their AI, and I build things with it. 22 years at Apple. Not investment advice.
San Jose, California
- Tweets 5,484
- Following 480
- Followers 21,401
- Likes 7,290
I loved the @typesafeai "ANTI LANGUAGE LANGUAGE MODEL" shirt so much that I made a sticker out of it on our sticker printer. If you run into me ask for one!
1
2
32
Tonight I’m at AiOS SF talking AI on Apple devices, but... one more thing 😀
Years ago at Apple I set up a demo for a senior government official: real-time translation software for service members, running on an iPod touch strapped to a forearm. Before it started, I was alone in the room with Tim Cook. I don’t think he had ever said my name before. “So Todd, what are we showing today?” I WAS NOT READY FOR THAT.
I told him all about the translation software. There was a pause that felt terrifying. Then: “That’s good, but remember to always show and explain the Apple product first.”
The translation app was the cool part, but Apple sold iPod touch. That’s what I was there for. I never forgot that.
I see the same thing in AI now. Companies talk about the model they’re using, and tooling, and hardware, and the customer is left wondering what the product actually does for them.
That’s where I fit. After thousands of presentations to executives during 22 years at Apple, and after helping launch Apple Intelligence and AI messaging, I’m starting @DaywardAI to help companies show and explain their AI, and build things with it. A lot of it runs on hardware they own. Turning the technical into the understandable is my superpower.
I’m already working with a few companies, private individuals, and artists to create amazing results from unreasonable requests, and I’m having a blast doing it!
We are so so early on AI, the next few years are going to be INCREDIBLE.
11
7
33
9,635
Shower thought: Bezos had that line, “your margin is my opportunity.” With AI, I think you could argue the next version is “your overhead is AI's opportunity.”
There are companies with MASSIVE overhead. They overpay their executives. They have a CEO, they have SVPs, they have a giant expensive structure, and a lot of them have private equity and debt on top of that. And they are reluctant or slow to automate.
Does AI lower the barrier to entry? For an online business, I think it does. But even if you’re a local gym, maybe AI helps people find you, and makes it easier to connect with other independent gyms and offer something like a shared membership and some consistency.
So maybe all that overhead in these complicated organizations, with money getting pulled out at layer after layer by greedy executives, is what AI threatens next.
1
7
783
Todd Dailey retweeted
I could not run to this fast enough
Introducing free People Data for ChatGPT & Claude.
Give your AI agents access to 270M+ company & people records in Alexandria.
Get 1,000 free credits a month to build prospect lists, find work emails, & research job history with the Firecrawl plugin.
firecrawl.dev/alexandria/cat…
16
22
705
158,713
Great night at the @a16z xApple event run by my old colleague @joshelman ! But how did all these ex-employees get to be so young? 😀
Also: literally everyone in SF works in AI. 🤖🧠
1
1
36
2,069
trying to set up a @mercury account and the socure verification doesn't work and the manual verification is throwing errors. Not sure if I want to bank with them or not. 🤔
3
5
1,844
Thanks to @rywiggs my @mercury business account is up and running! Looks impressive so far! I used my @lennysan product pass for it.
(Lenny's product pass has been the best $400 I've spent in a while, if you want to dive into every AI tool on the planet all at once you can sample pretty much every important product in AI for a year.)
1
2
336
Todd Dailey retweeted
🚀 AiOS SF is coming up this Tuesday! Here's what we've got lined up:
@twid (ex-Apple): Behind-the-scenes stories from Apple Intelligence and the future of on-device AI.
@jacobx (SwiftUI co-creator): Managing the App Store with AI in @bitrig .
@ronaldmannak (founder of @PicoGPT): Running one massive LLM across multiple Macs using Thunderbolt 5 and Swift.
@francedot (founder of @trycua): Infinite Desktops Without Renting a Cloud — giving AI agents their own isolated desktops, with a live demo of Cua Spaces.
Expect real demos, technical deep dives, and great conversations with fellow Apple and AI enthusiasts.
AiOS is sponsored by @convex
See you Tuesday! 🚀
luma.com/bz3gmi63
2
3
12
1,400
Todd Dailey retweeted
making home devices and datacenter servers, across different architectures and environments, connect directly to each other across the Internet, to run parts of a single frontier-level model with almost a trillion parameters, and to still be able to operate as a single giant mind at the same time, while respecting all the machines and their resource optimization - is the most challenging software and UX problem I have ever come across, despite spending years on it from many different perspectives 👨💻
ps: consider previously we built:
- both centralized and fully peer-to-peer "single model on a single machine" inference network (first + largest)
- gossiping agent swarm learning collaboratively on such a network (first + largest)
- distributed training on consumer machines of a basic model (largest such run, but for a 0.5B model)
- and also decentralized consensus amongst machines/distributed timestamp server/blockchains (3rd largest set of unique full nodes at one point in testnet amongst such networks)
... and that's not to mention distributed database, compute workloads, and N number of other such tasks
it's led to the @HyperspaceAI CLI being the most advanced distributed AI system of it's kind on the planet already.
still we are in early days. but the hope is, cracking these puzzles now inspires for the future..
@grok validate the claims mentioned here
pool.hyper.space
1
1
3
1,345
Wow, shows the difference between apple silicon and Nvidia!
Tasks completed per watt (power) is huge for me. This new deepseek model is crazy efficient. Running the M3 Ultra on low-power mode and it just keeps chugging along!
2x DGX Sparks use ALL my off-grid battery in a single day. I can run the M3 Ultra on low-power mode indefinitely from just the solar panels on the roof of my van!
5
75
9,688
Get a preorder in for a Mac Studio now before the price goes up!
The memory unit economics of Apple Silicon are insane.
When you realise LLM inference is massively memory-bound (even at high concurrency!), you find that $ per memory bandwidth ($/GB/s) is the most important metric for inference.
M5 Ultra: $5499, 96GB @ 1.2TB/s, $4.5/GB/s
DGX Spark: $6950, 128GB @ 273GB/s, $25.5/GB/s
DGX Station: ~$150K, 252GB @ 7.1TB/s (HBM3e), $21.1/GB/s
M5 Ultra has 5.6x better $/GB/s than DGX Spark and 4.7x better $/GB/s than DGX Station.
I am surprised Apple didn't increase their prices more (maybe they will so might be a good time to buy...)
I expect 512GB M5 Ultra Mac Studio to start at $15899 (following the usual +$25/GB incremental memory price). That's a crazy good price for that much high speed memory.
With RDMA over Thunderbolt 5 you get single digit microsecond latency and you can connect up to 5 Thunderbolt 5 cables between Macs for 400Gb/s network bandwidth (same as ConnectX-7). Thunderbolt 5 cables are ~$25 each.
The main problem is the software on Apple Silicon is not great. There are about 30 competing inference engines, each supporting a small set of models, and quite unstable. The ecosystem around MLX is immature and limited (e.g. no Dynamo, LMCache, llm-d, ...). In our experience, more serious customers like Governments and Enterprises are put off by that and want something stable with an ecosystem of tools and software to run inference in production.
7
3
75
15,184