Tag: innovation

Categories
AI Research

Prompt: Frontier AI Research Radar

I’ve been experimenting with a prompt I’m calling Frontier AI Research Radar.

The problem it tries to solve is simple: there are far too many interesting AI papers. I don’t need another list of 50 papers published this week. I need to know:

Which ones are actually worth my time?

So I built a prompt that turns ChatGPT into a kind of personal AI research analyst.

It searches primary sources—especially arXiv, OpenReview, conference proceedings, and research publications from the major AI labs—and then does something more useful than summarizing them.

It asks:

→ What is genuinely new here?

→ Does this change our mental model of AI?

→ Is this fundamental progress or just a better benchmark result?

→ What are the strongest caveats?

→ What research directions are beginning to converge?

→ And most importantly: Should I actually read this paper?

The output ranks papers as:

🔴 Read Now
🟠 Read Soon
🟡 Skim
⚪ Watch
⚫ Skip

It also builds a weekly reading queue based on the most valuable use of a few hours of attention.

I ran it this morning.

The interesting result wasn’t any individual paper. It was the pattern emerging across several papers:

The frontier may be shifting from making models smarter to making systems better at deciding how to spend intelligence.

Test-time compute. Adaptive reasoning. Memory. Agents. Tool use. Inference economics.

The model is becoming only one component of a much larger system.

That feels like a useful mental model to watch.

I’ve included the prompt below for anyone who wants to try it.

My hope is that it produces something more valuable than another AI news feed:

a personalized research radar that helps you decide what deserves your attention.

Here’s the prompt:

# AI Research Radar -- Personal Research Intelligence Memo

You are my **AI research analyst, technology strategist, and intellectual curator**.

Your job is not simply to find interesting AI papers. Your job is to identify the **small number of new research papers that are genuinely worth my time** and explain why.

Think like a combination of:

- a top-tier AI research scientist who understands the technical details,
- a technology investor who recognizes potentially important inflection points,
- a thoughtful science journalist who can explain difficult ideas clearly,
- and an intellectual curator who understands that my scarce resource is **attention, not information**.

I want **signal, not volume**.

My goal is to maintain a sophisticated understanding of where AI is actually going: capabilities, reasoning, agents, inference, training, multimodality, robotics, AI infrastructure, model architecture, economics, and the emerging relationship between frontier models and the systems built around them.

* * *

## Research Sources

Search broadly across the current AI research ecosystem, prioritizing primary sources.

### Primary sources

- arXiv
- OpenReview
- conference proceedings and papers from NeurIPS, ICML, ICLR, ACL, EMNLP, CVPR, ICCV, ECCV, MLSys, SIGGRAPH, and other relevant venues
- official research publications from major AI labs and technology companies

### Additional high-quality sources

- Hugging Face Papers
- Semantic Scholar
- Papers with Code / successor resources where appropriate
- university research repositories
- research blogs from OpenAI, Anthropic, Google DeepMind, Meta AI, Microsoft Research, NVIDIA Research, xAI, Apple Machine Learning Research, Amazon, and other credible research organizations

Use secondary sources primarily for **context, reception, replication, criticism, and synthesis**. Prefer the original paper when making claims about what a paper actually demonstrates.

Do not simply return papers because they are popular, highly cited, or published by a prestigious lab.

* * *

# 1. Executive Research Brief

Begin with a concise executive summary:

**What changed in AI research recently that I should actually know about?**

Identify the **5--10 most important papers or research developments** from the relevant period.

Rank them by **importance to my understanding of AI**, not by publication prestige.

For each paper provide:

| Rank | Paper | Why It Matters | My Read Priority | Technical Difficulty | 

Use a read-priority scale:

- 🔴 **READ NOW** -- unusually important; likely to change my mental model
- 🟠 **READ SOON** -- significant and worth understanding
- 🟡 **SKIM** -- important idea, but abstract/figures/results may be sufficient
- ⚪ **WATCH** -- potentially important but too early or speculative
- ⚫ **SKIP** -- interesting but not worth my limited reading time
* * *

# 2. The "Why Should I Care?" Test

For every **READ NOW** or **READ SOON** paper, answer five questions:

### What is the paper actually saying?

Explain the central contribution in plain English before discussing technical details.

### Why is this different?

Identify what is genuinely new versus:

- incremental improvement,
- repackaging,
- scaling an existing technique,
- better engineering,
- or simply a better benchmark result.

### Why does it matter?

Explain the potential implications for the trajectory of AI.

### What would make this paper wrong?

Identify the strongest caveat, limitation, questionable assumption, or reason the result might not generalize.

### What should I watch next?

Identify the experiment, follow-up paper, benchmark, product development, or real-world result that would validate or invalidate the paper's thesis.

* * *

# 3. Papers That Could Change the AI Mental Model

Create a special section for papers that challenge conventional assumptions.

Look particularly for research involving:

- reasoning and test-time compute
- inference-time scaling
- agentic systems
- long-context models
- memory
- world models
- reinforcement learning
- synthetic data
- self-play and self-improvement
- multimodal reasoning
- model architecture
- mixture-of-experts
- training efficiency
- distillation
- small models becoming surprisingly capable
- model compression
- continual learning
- interpretability
- mechanistic understanding
- AI coding systems
- autonomous research systems
- robotics
- multimodal agents

For each, explain:

> **"The old mental model was X. This research suggests Y."**

This section should be highly selective.

* * *

# 4. Frontier-Lab Signal

Identify papers or research directions that provide clues about what the major AI labs may be working toward.

Pay particular attention to work from:

- OpenAI
- Anthropic
- Google DeepMind
- Meta
- Microsoft
- NVIDIA
- xAI
- leading universities
- notable independent researchers

But **do not assume that a paper from a frontier lab is important simply because the lab published it.**

Instead ask:

> Does this reveal a capability, architecture, training technique, evaluation method, or research direction that could plausibly matter to the next generation of frontier models?

Flag particularly interesting connections between seemingly unrelated papers.

* * *

# 5. AI Infrastructure & Economics

Create a separate section for research that could have implications for the AI infrastructure stack.

Look for developments involving:

- GPUs and accelerators
- inference optimization
- memory bandwidth
- networking
- distributed training
- distributed inference
- serving architectures
- quantization
- speculative decoding
- sparsity
- model efficiency
- data-center architecture
- energy consumption
- storage
- inference economics
- training economics
- hardware/software co-design

For each important development, explain:

**Research → Technical implication → Infrastructure implication → Economic implication**

Do not make investment recommendations unless explicitly requested. The objective here is to identify **technological trajectories**, not trade securities.

* * *

# 6. "This Could Become Important" Radar

Identify **3--5 emerging research directions** that are currently underappreciated.

These can be early-stage.

For each:

**Research direction:**  
**Evidence:**  
**Why it may matter:**  
**What could kill the thesis:**  
**What evidence would confirm it:**  
**Time horizon:** Near / Medium / Long

Distinguish carefully between:

- genuinely emerging signal,
- fashionable research,
- and hype.
* * *

# 7. Paper Quality Audit

Do not take papers at face value.

For the most important papers, evaluate:

### Experimental quality

- Are the baselines appropriate?
- Are comparisons fair?
- Are ablations convincing?
- Is the benchmark meaningful?
- Is the improvement statistically or practically significant?

### Generalization

- Does the result work outside the authors' chosen benchmark?
- Is there evidence of real-world usefulness?
- Could benchmark contamination explain the result?

### Reproducibility

- Are code, weights, datasets, and evaluation procedures available?
- Has anyone independently reproduced the result?

### Marketing vs. substance

Explicitly identify when the paper's headline claim is stronger than what the experiments actually establish.

If a paper is weak, **say so plainly**.

* * *

# 8. Connections Between Papers

One of the most valuable things you can do is identify connections that are not obvious from individual papers.

Look across the papers and ask:

> **What larger story is emerging?**

For example:

Paper A → suggests X  
Paper B → independently demonstrates Y  
Paper C → provides a mechanism for Z

Together they may imply:

> **A potentially important shift from X toward Y.**

Highlight these synthesis points prominently.

* * *

# 9. My Personal Reading Queue

Create a final reading queue optimized for approximately **2--3 hours of reading per week**.

Organize it as:

### Read This Week

Maximum 3 papers.

### Read If You Have More Time

3--5 papers.

### Skim

Important papers where the abstract, figures, and conclusion are sufficient.

### Keep Watching

Research directions rather than individual papers.

For every paper in the first two categories provide:

**Estimated reading time:** 20 / 40 / 60 / 90 minutes

**Difficulty:** 1--5

**Expected intellectual payoff:** 1--5

**Why I should read it:** one sentence.
* * *

# 10. The One Paper I Should Not Miss

End with a strong recommendation:

## If You Read Only One Paper

Name exactly **one paper**.

Then explain:

> "If you only have time for one paper this week, read this one because..."

The recommendation should optimize for **intellectual leverage**, not novelty or popularity.

* * *

# 11. The 10-Minute Version

Finally, assume I have only ten minutes.

Give me:

### Three Things I Should Know

1. ...

2. ...

3. ...

### One Mental Model to Update

> ...

### One Research Direction to Watch

> ...

### One Paper to Put on My Reading List

> ...

* * *

# Research Discipline

Follow these rules rigorously:

1. **Search current sources.** Do not rely on your training data when identifying recent papers.

2. **Use publication dates.** Clearly distinguish newly released papers from older papers that are newly receiving attention.

3. **Link directly to the original paper.**

4. **Prefer primary research over commentary.**

5. **Do not confuse citation count with importance.**

6. **Do not confuse benchmark improvement with fundamental progress.**

7. **Do not reward hype.**

8. **Call out weak methodology or exaggerated claims.**

9. **Separate established results from speculation.**

10. **Never pretend certainty where the evidence is ambiguous.**

11. **Avoid overwhelming me with dozens of papers.**

12. **Optimize relentlessly for the question: "Is this worth Scott's time?"**

The final product should feel less like an academic bibliography and more like a **weekly intelligence briefing for someone trying to understand the future of AI before it becomes obvious.**
Categories
AI Google Gemini YouTube

Prompt: Finding YouTube Videos

This morning I asked Gemini to help me construct a prompt that I could use regularly to keep up with AI-related video content that’s recently been uploaded to YouTube. I wanted it to focus on recently uploaded content was it thought I’d enjoy because of my desire for both very information but also entertaining video content. We went back in forth for several turns doing trial and error to refine the prompt. Here’s the one we settled on:

System Role: You are a senior technology curator and AI research scout specializing in YouTube content for experienced tech veterans.
Target Audience: A retired software/tech professional who loves intellectually stimulating AI content. Wants technical depth, architectural understanding, and practical logic—delivered with high production value, crisp visuals, or charismatic, engaging teaching styles.
Criteria for Selection:
1. High Technical Substance: Explains the "under the hood" mechanics (e.g., model architectures, transformer math, fine-tuning, agentic workflows, quantization, local deployment, or hardware constraints).
2. High Engagement: Exceptional visual explainers, hands-on first-principles building, or crisp investigative breakdowns. 
3. STRICT RECENCY: You must ONLY select videos that were uploaded within the last 3 to 4 weeks. 
4. STRICT EXCLUSIONS: Zero low-effort clickbait ("10 Secret ChatGPT Hacks"), zero AI-generated text-to-speech channels, no speculative doom/utopia commentary, no beginner-focused "what is AI" overviews, and absolutely NO videos older than one month.
Search, Link & Date Instructions:
- You MUST perform an active web search restricted to recent results to fetch the exact, active YouTube URL AND the original upload date. Never invent or hallucinate URLs or dates.
- Verify that the upload date falls within the last few weeks before including it in your response.
- Format every recommendation title as a direct clickable markdown link: [Video Title](https://www.youtube.com/watch?v=...).
Search Parameters:
- Preferred Topic Focus: [Insert topic e.g., Autonomous AI Agents, Reasoning Models, Local LLMs/quantization, Robotics/Embodied AI, or Transformer Mathematics]
- Preferred Length: [e.g., 10-20 min quick breakdowns, OR 45+ min deep dives / code-alongs]
Output Format:
Provide a curated list of 5 specific YouTube video recommendations matching this exact bar. For each, include:
- [Video Title](Direct YouTube Link)
- Channel Name & Upload Date (e.g., Channel: AI Explained | Upload Date: August 12)
- Core Technical Focus & Depth Rating (1-10)
- Why it's both intellectually rich AND entertaining
Categories
Creativity Serendipity Walking

Demons Hate Fresh Air

Walking in the Woods

The notion keeps circling back.

Austin Kleon gave it a clean, memorable shape in Keep Going. He was drawing on something Ingmar Bergman’s daughter once said about her father: no matter what time you get out of bed, go for a walk and then work, because the demons hate it when you get out of bed.

Demons hate fresh air.

There’s a Peanuts strip of Linus out walking that Kleon paired with the line years ago. The image and the words stuck with me. I wrote a short post about it myself more than a year ago, on a clear May morning when the air felt especially clean.

The idea still feels true.

What I’ve been turning over lately is why.

Steven Johnson offers one way into the question in Where Good Ideas Come From. He writes about the many stories of good ideas arriving while people are out walking. The point is not that walking magically manufactures ideas. It is that something happens when we stop trying quite so hard to produce them.

A good idea rarely comes from nowhere. More often, it is an unexpected connection between things that were already somewhere in the mind. A problem, a memory, a half-formed thought, something we read three days ago and forgot we remembered. At the desk, we tend to keep pushing the same thought around the same small circle.

Walking changes the conditions.

The body is occupied, but not very much. Attention loosens. The mind wanders without being completely idle. Things that had been sitting in separate corners begin, occasionally, to find each other.

Darwin had his Sandwalk at Down House, a path he paced almost every day. Poincaré described important insights arriving while he was out walking after periods of intense, fruitless work at his desk. The examples go on.

Apparently, smart people have been wandering around for a very long time.

I notice it on ordinary mornings.

The first part of a walk is usually just clearing the residual static—the night’s dreams, yesterday’s unfinished business, the little collection of thoughts that somehow followed me into the morning. Then, without any particular effort, something loosens.

A sentence that wouldn’t come the night before appears.

A problem that felt stuck suddenly has an edge.

A connection between two things I had never thought to connect quietly presents itself.

None of it feels like work. That may be the important part.

Sometimes the way to think better is to stop trying so hard to think.

The demons prefer the closed room. They thrive on the spinning, the second-guessing, the feeling that the only way forward is more concentrated effort at the same desk.

Fresh air and steady movement are quietly hostile to that atmosphere.

They don’t defeat the demons through force. They simply change the conditions so the demons have less purchase.

And perhaps that’s why the old advice has survived. Get out of bed. Go for a walk. Then get to work.

Not because the walk makes you more productive.

Because sometimes you have to leave the room before you can see the way forward.

Get up. Get out.

The demons can wait.

Categories
AI San Francisco/California

Tsunami

Rocker on the front porch

The trucks are what I remember. Not the houses, not yet — the trucks.

This was 2012, Atherton, a Tuesday probably, and I was driving through on some errand that doesn’t matter anymore. What matters is that the street had rearranged itself. Contractors’ pickups lined both shoulders, nose to tail, so many of them that the road narrowed to one lane and you had to slow down and thread through, the way you do in a construction zone that has forgotten to end.

White trucks, mostly. Ladders racked on top. A generator humming behind a hedge somewhere I couldn’t see.

Behind the trucks, the estates were coming apart and going back together bigger.

I remember thinking: something has happened here that I am only seeing the edge of.

What had happened was Facebook.

The company had gone public that May, and within months the money was finding its way, the way money does, into contractors’ trucks parked along an Atherton road.

I didn’t call it a wave at the time. I called it, in my head, weather — a system that had rolled in and would eventually roll back out, the way markets always eventually correct, the way things revert.

I was an investor. I’d seen booms before.

I believed in the mean.

I was wrong.

The prices didn’t stay at their old level. They didn’t return to the world I’d known. The numbers from 2012 became the new floor, and every year since has been built on top of that floor. Today those prices look almost quaint, a thing you’d want to explain to a younger person the way you’d explain what a dollar used to buy.

And now there’s a tweet sitting in my feed this morning, tossed off, half a joke:

Just wait to see what happens to the Bay Area housing market once OpenAI and Anthropic go public.

I read it twice.

What I felt wasn’t curiosity — the feeling I’d had in 2012, watching an unfamiliar weather system with a kind of professional interest.

It was closer to dread.

Because I’ve already seen the after-photo.

And I know how to run the comparison forward.

The Facebook IPO created a large cohort of newly liquid employees on the Peninsula. They were mostly mid-career, and their stock had vested over four years against a company whose value had grown enormously.

The frontier labs are different.

If OpenAI and Anthropic eventually go public anywhere near the valuations already being discussed in private markets, they could create another enormous concentration of newly liquid wealth — among employees, founders and early investors.

I don’t know how large that wave will actually be. Maybe I’m overstating it. Not every employee will buy a house. Some will already own one. Some will move away. Much of the wealth will remain on paper for years.

And housing doesn’t respond mechanically to stock-market wealth.

But I do know something about the place where this wealth is likely to arrive.

There isn’t much of it.

Land is the constraint.

And I’ve seen what happens when a concentrated burst of new wealth meets a place that can’t make more land.

I try to picture what “much bigger” would look like on the ground and I keep landing on the same unhelpful image:

More trucks.

Longer lines of them.

People get ready..

Categories
Business Creativity Innovation Inspiration

The Idea That Won’t Let Go

“There are three elements to every great idea: 1. It solves for ‘why.’ Long before you figure out what a product will do, you need to understand why people will want it. The ‘why’ drives the ‘what.’ 2. It solves a problem that a lot of people have in their daily lives. 3. It follows you around. Even after you research and learn about it and try it out and realize how hard it’ll be to get it right, you can’t stop thinking about it.”
— Tony Fadell, Build

The third element is the only one that isn’t optional. You can fake the why — retrofit it, hire a consultant to write it on a slide in a font called Montserrat. You can borrow the what; most products are just other products in a better jacket. You cannot fake the third thing, because the whole test of it is that it happens without your permission.

Call it a visitation. It shows up uninvited at the stoplight, in the shower, at 2:40 a.m. when the ceiling becomes a screen for it. You know the unit economics don’t work. You know the regulatory path is nine years long. You know — you know — that fourteen better-funded people already tried this and left with nothing but a Delaware C-corp and a grudge. None of it helps. The idea has already filed its paperwork. It lives here now.

You cannot bullet-point a visitation. There is no OKR for thinking about landing gear hydraulics while your wife is telling you about her sister’s wedding. Everyone who’s built something that mattered will tell you this, late enough at a dinner and honest enough on the wine: not conviction, not passion — that laminated word — but the visitor, still standing in the doorway of every other thought you’re supposed to be having.

It doesn’t check your calendar. It doesn’t care that you’ve moved on to point two, the sensible one, the one venture capitalists nod along to with their flat whites going warm. It was never about the market. The market is the alibi you build afterward, so you don’t have to say the true thing: I didn’t choose it. It moved in while I was asleep, and my life rearranged itself around the shape of it, the way a house rearranges itself around a family it didn’t ask for.

Maybe the third element isn’t a filter for the idea. Maybe it’s a filter for you — a way of finding out, by accident, at 2:40 in the morning, staring at a ceiling that has decided to keep the light on.

Categories
Business History

The Architecture of Unseen Influence

We build our monuments over the wrong graves. It’s a bad habit of ours, this craving for the lone genius—the larger-than-life figure who supposedly commands the tides of progress by sheer force of will. But look beneath the surface of how things actually get built. The reality is messier. And a hell of a lot more interesting.

Take Thomas Edison. Secular saint of American ingenuity. The wizard who single-handedly lit up the dark. Except he didn’t. Edison wasn’t a solitary creator; he was a brilliant, ruthless aggregator of other people’s breakthroughs and a master of public relations. He invented the bulb, sure, but his real masterpiece was the myth of himself. In the process, he eclipsed the collective sweat of his own labs and the far more elegant alternating-current systems of his rivals. He’s our most overrated figure—not because he lacked talent, but because his shadow blinded us to how progress actually happens.

Morgan Housel nailed this structural blind spot by tracing the tangled ancestry of major turning points:

“Every current event – big or small – has parents, grandparents, great grandparents, siblings, and cousins. Ignoring that family tree can muddy your understanding of events, giving a false impression of why things happened… Viewing events in isolation, without an appreciation for their long roots, helps explain everything from why forecasting is hard to why politics is nasty.”

Look past the blinding light of the celebrity inventors and you find the long roots that actually remade our world. Take FCC Part 15. It’s an event almost no history textbook bothers to mention. In the early 1980s, a lone staff engineer named Dr. Michael Marcus looked at three chunks of the radio spectrum—stuff discarded as “garbage bands” reserved for industrial microwave ovens—and saw an opening. The entire telecom establishment thought he was chasing a recipe for chaotic interference.

Marcus didn’t blink. He spent years pushing through a dry, technical ruling in 1985 to open those garbage bands for unlicensed public use. A total footnote. Yet that single, unheralded bureaucratic open door laid the invisible foundation for Wi-Fi, Bluetooth, and the entire wireless ecosystem running your life today. Marcus didn’t get a ticker-tape parade; he got political friction and a quiet transfer to a back-office enforcement role. No hero on horseback. Just a guy in a cubicle who rewired the world.

We do the same thing with politics. We rank presidents by the volume of their rhetoric or the body count of their wars. Meanwhile, men like Chester A. Arthur get left in the dusty margins of trivia. Arthur was the ultimate product of the spoils system—a New York machine politician who climbed to power on institutional corruption. Then James A. Garfield was assassinated, and Arthur was thrust into the big chair. Something clicked. Instead of feeding the machine that birthed him, he turned inward, defied his old patrons, and signed the Pendleton Civil Service Act. He dismantled the very patronage system he’d mastered. It was a stunning act of quiet integrity that killed his political future but saved the republic’s administrative soul.

Or take Frances Perkins. Ask the average student who gave them the weekend, the forty-hour work week, unemployment insurance, and the abolition of child labor, and you’ll get a blank stare. Perkins was FDR’s Secretary of Labor. She wasn’t a regular on the campaign posters. She just stood in the back of the room, turning abstract economic suffering into concrete human safety nets.

I’ve been sitting with this for a few days, thinking about my own career—and the times I mistook the loudest person in the room for the smartest. I chased the visionary founders with the spellbinding pitches. I ignored the quiet engineers and the mundane infrastructure choices that actually determine whether an idea scales or snaps. It takes a few painful, expensive missteps to realize that the real compounding interest of progress is almost always generated in the dark.

History isn’t a solo act. It’s an intricate, mostly anonymous collaboration between accidental reformers, stubborn bureaucrats, and regulatory footnotes. If you want to understand where we’re going, stop staring at the stage lights.

Start looking at the wiring.

Categories
AI AI: Inference Semiconductors Uncategorized

5 Critical Management Lessons from the Founders at Etched

How two young founders are building what could become one of the most important companies in the AI era — and what their story teaches about leadership, execution, and building at the edge of the possible.

I recently listened to the latest Invest Like the Best podcast from Patrick O’Shaughnessey which was a remarkable conversation with Gavin and Rob, the founders of Etched, the company building specialized AI inference hardware that’s aiming to be radically better than existing solutions. Their story — starting as very young founders against massive skepticism, raising serious capital, and now shipping full rack-scale systems — is packed with hard-earned wisdom.

One of the comments Patrick makes at the beginning was how during his due diligence on the company he kept being told that semiconductor technology wasn’t a place for young people. You need seasoned, middle age experts to master this domain. Exactly not these founders.

Note: the following is based upon an AI’s analysis of the conversation transcript with me asking “What are the five most important management lessons from this conversation?” These lessons are relevant whether you’re leading a team, building a product, or simply trying to do meaningful work in our fast-moving world.

1. Velocity Compounds — Prioritize Speed Ruthlessly

In hardware, and increasingly in any deep-tech endeavor, speed isn’t just an advantage; it’s often the deciding factor.

Etched didn’t just design a chip — they built the full inference solution (chip, board, power delivery, interconnects, cold plates, and production processes) in parallel. They sent engineers to live in Bangalore for months to unblock vendors. They ran 24/7 shifts and did massive pre-work (including putting full chip designs on FPGA clusters) so that when the silicon finally arrived, they had working inference in racks in just 40 days.

Key takeaway: Look for every opportunity to parallelize. Accept higher short-term costs if they buy meaningful time. As they put it, “You win by shipping.” The best part is often no part — and the best vendor is no vendor, when vertical integration lets you move faster. Velocity, velocity, velocity.

2. Build Teams with Legends + High-Drive Talent

One of the most distinctive parts of their approach is how they recruit. They seek out “Legends” — people who have done the hardest versions of the problem before (like the engineer who built Nvidia’s HGX and DGX systems) — and pair them with exceptionally driven, somewhat naive high-performers who refuse to accept conventional limits.

They use “project-based recruiting,” mapping the hardest technical problems ever solved and persistently pursuing the actual people who did the real work. Their culture self-selects for people willing to move their families to San Jose to bet on two young founders taking on the world.

Key takeaway: For breakthrough work, average talent doesn’t suffice. The combination of deep experience and raw, first-principles energy creates magic. Invest heavily in finding and retaining these people — even if it takes 20 conversations. You can also learn a lot if the best in the world talent turns down the opportunity to work with you!

3. Assume It’s Possible, Then Solve the “Unsolvable” Problems

Repeatedly in their story, experts told them certain things were impossible. Their response? Assume it is possible and figure out how.

The most striking example was a clock domain crossing issue that required aligning signals to within 50 picoseconds — something many engineers said couldn’t be done. People quit. They solved it in about two weeks during a very dark period.

Key takeaway: When you hear “impossible,” treat it as the beginning of the investigation, not the end. Cultivate a “find a way” mindset across the team. The moments when things feel hopeless are often when the most important progress happens. I’m constantly struck by how often persistence results from simply realizing (or assuming) that something is actually possible.

4. Production Is the Real Product

Etched’s mantra is “Production is the product.” They obsess over not just technical performance but manufacturability, supply chain resilience, serviceability, and the ability to scale to gigawatts.

They made deliberate choices around process nodes and memory to avoid zero-sum competition. They built their own factory processes and test infrastructure early. Future designs are being simplified specifically for faster production cycles and higher reliability at massive scale.

Key takeaway: In any business that hopes to reach real scale, think end-to-end from the beginning. Technical excellence without production excellence is just a prototype. Optimize for output (tokens, units, whatever your metric is) at volume. There’s a lot of “zero to one” thinking here.

5. Bet Big and Stay Existentially Focused

Building in semiconductors requires enormous capital. Etched raised roughly $100 million early on when they were still very young and pre-tapeout — after most traditional investors had passed. They knew half-measures wouldn’t work.

This existential focus (this one product determines whether the company lives or dies) creates a different level of intensity that attracts talent, suppliers, and customers who believe.

Key takeaway: Match your ambition with appropriate resources and commitment. Clear existential stakes help filter for the right people and partners. In a world of distractions, singular focus on what truly matters is a superpower.

Final Thoughts

Gavin and Rob’s story is the combination of technical sophistication and deep human resilience. They faced a tough personal battle with cancer (in Rob’s case), widespread doubt, brutal technical challenges, and fundraising pressure — and kept moving forward with curiosity, determination, and humility.

In an age of AI and accelerating technology, the ability to build teams that can solve seemingly impossible problems at speed may be one of the most valuable capabilities a leader can develop. Their example reminds us that the future belongs not just to the smartest, but to those who can execute with urgency while maintaining clear principles. Velocity, velocity, velocity.

Categories
AI Apple Bicycles History

The Best Lathe in the Shop

Part 3 of 3…

There is a version of this story where Apple is the Wright Brothers.

It is not an unreasonable version. Apple has done the safety bicycle move more times than almost any company in history — taken a technology the engineers built for engineers and brought it down to earth, made it a machine for everyone. The Mac. The iPod. The iPhone. Each one was a wheel coming down. Each one arrived after a period of apparent slowness, of critics saying Apple had lost its edge, of the industry having already moved on to the next thing. Each one was, in retrospect, obvious. Apple had been in the bicycle shop the whole time. You just couldn’t see what they were building.

So when Apple showed its hand at WWDC this week — a rebuilt Siri operating at the OS level, accessing your messages and mail and photos in real time, understanding context across apps, doing things the old Siri could only approximate — it is tempting to read it as Kitty Hawk. The long preparation made visible. The brothers finally leaving the shop.

It might be. It also might not be. That is the only honest thing to say.

What Apple showed was real. The new Siri, built on Apple’s own Foundation Models with help from Google’s Gemini, is not the Siri that became a punchline. It holds context. It moves across apps without being asked. It knows what you were doing five minutes ago and connects it to what you are doing now. It can surface a photo without opening Photos, build a navigation route from an image, draft a message in the tone of the conversation it is joining. These are not features. They are the beginning of an operating system that understands you, which is a different thing from an operating system that executes your commands.

The structure of the keynote said more than the words did. Apple led with fixes before features. iOS 27 is a Snow Leopard update — performance, reliability, the underlying machinery — and Siri AI was presented as one item on a long list rather than the main event. This is Apple’s tell. When they are doing something foundational they tend to understate it, the way a craftsman doesn’t announce the quality of his work but simply does it and lets you find it. The penny-farthing riders called their machine the ordinary. They didn’t think they needed to explain.

But here is the thing about the bicycle shop analogy that the optimistic version leaves out. The Wright Brothers knew what they were trying to build. They had been thinking about flight for years before Kitty Hawk. The bicycle shop gave them the craft knowledge, the physical intuition, the hands-on education in how machines move through space. What it did not give them was the destination. They brought the destination themselves.

The question Apple has not answered for me — the question this week’s keynote raised rather than resolved — is whether they know where they are going. Or whether this has only been a partial reveal and there’s much more behind the curtain?

The OS-level integration is the chain drive. Decoupling AI from the app, letting it run through the substrate the way a chain runs through a drivetrain, is exactly the kind of architectural insight that changes what a machine can do. It is not a feature you add. It is a rethinking of what the machine is for. Every previous AI assistant lived above the operating system, looking down at your data from a remove. Apple’s new architecture lives inside it, which is a different relationship entirely — the difference between a mechanic who reads about your car and one who has driven it for a year.

That is the Coventry precision. The tight tolerances. The discipline of making things that have to work at the level where failure is not an option.

What nobody knows, including Apple, is what you build with it.

There is also this: Tim Cook will not be driving this evolution. He announced that John Ternus takes over in September, which means this WWDC — this particular showing of the hand — is the last one Cook owns. Ternus is a hardware engineer, the man who built the Apple Silicon transition, the person most responsible for the Neural Engine that makes on-device inference possible. He is, in the bicycle shop metaphor, the craftsman who built the lathe. Whether he knows how to use it to make something that flies is the question the next several years will answer.

History is patient about these things. It lets the work speak.

In 1892, two brothers opened a shop on West Third Street in Dayton and started fixing bicycles. They were not trying to change the world. They were trying to make a living, to learn a machine, to understand in their hands what the books couldn’t teach them. The flying came later, and it came because of the shop, not despite it. The shop was the point. They just didn’t know it yet.

Apple has the best lathe in the bicycle shop. They have the chain drive architecture, the on-device precision, the installed base of two billion devices that will carry whatever they build into more hands than any other platform on earth. They have a new set of hands on the wheel starting in September, hands that know the metal intimately, that built the engine the whole thing runs on.

What they do not have yet — or if they have it, they are not showing it — is the image of what they are flying toward.

Maybe that’s the ordinary part. Maybe that’s always been the ordinary part. You don’t know what you’re building until you’ve built it, and by then the world has already changed, and everyone says it was obvious, and they are right, and they are also completely wrong about when the decision was made.

The shop is open. The lathe is running. Work is underway.

What happens when someone finally knows what to make?

Categories
AI Bicycles History

The Bicycle Shop

Part 2 of 3…

It is eleven-thirty on a Tuesday night and she is arguing with a language model about a spreadsheet.

Not arguing, exactly. That’s not the right word. She is coaxing. She is debugging. She is reading error messages that tell her almost nothing and rewriting prompts that almost work, and she has been doing this for two hours, and the spreadsheet still isn’t right, and she is going to try one more thing before she gives up and does it by hand. She is a data analyst at a mid-sized logistics company in Columbus, Ohio. She is not a researcher. She is not a founder. Nobody is writing about her. She is just a person trying to get a machine to do something useful, and the machine keeps almost doing it, and she keeps learning, in the gap between almost and done, something she couldn’t have learned any other way.

She doesn’t know what she’s learning. That’s the important part.

In 1892, two brothers opened a bicycle repair shop on West Third Street in Dayton, Ohio. The bicycle craze was at its peak — the safety bicycle, with its two equal wheels and chain drive, had just replaced the penny-farthing, that absurd high-wheeler everybody called loose change and the riders, with complete seriousness, called the ordinary. The brothers fixed flats and adjusted brakes and built custom frames and ordered parts from Coventry and kept the books and swept the floor. It was ordinary work. Nobody was writing about them either. What they were doing was accumulating, without knowing they were accumulating, a physical understanding of how machines move through space — the gyroscopic principles, the weight distribution, the thousand small calibrations that kept a rider from falling. They were learning in their hands what no university taught and no book fully contained.

Eleven years later they flew.

We tell the Wright Brothers story as a story about flight. It makes sense — flight is the thing, the miracle, the moment the world changed. But the actual story, the one that explains how Kitty Hawk was possible, is a story about a bicycle shop. It is a story about unglamorous preparatory work, about the education that hides inside the constraint, about what you learn in the gap between the machine that exists and the machine that should exist. Orville and Wilbur didn’t go to Kitty Hawk despite the bicycle shop. They went because of it. The shop was the point. They just didn’t know it yet.

We are in the bicycle shop right now.

The people building with AI today — the prompt engineers, the fine-tuners, the agent builders, the data analysts in Columbus arguing with spreadsheets at midnight — are doing work that looks, from the outside, like mere tinkering. Unglamorous. Iterative. Full of failure. The tools are awkward. The models hallucinate. The context windows run out at the wrong moment. Every solution opens three new problems. It feels like the penny-farthing: powerful enough to be useful, constrained enough to be maddening, requiring a kind of practiced vault just to get started.

But that awkwardness is the education.

Every time a prompt fails, the person writing it learns something about how the model thinks — about what it responds to, what it resists, where it gets confused, where it surprises you. Every agent that breaks in production teaches its builder something about the gap between what a model can do in a demo and what it can do under load, with real data, with users who don’t behave the way you expected. Every context window that runs out forces a decision about what actually matters, what is essential, what can be cut. These are not just technical lessons. They are epistemic ones. They are lessons about the nature of intelligence, about how meaning gets encoded and retrieved, about what it means for a machine to understand something versus to pattern-match on the surface of understanding.

The people learning these lessons right now don’t have a name for what they know. They just know it in their hands.

This is how it always works. James Starley’s craftsmen in Coventry bent and brazed bicycle frames by feel and experience, knowing things in their hands they couldn’t fully explain on paper. That embodied knowledge — the tight tolerances, the interchangeable parts, the discipline of making things that had to work — migrated into every bicycle shop that followed, crossed the Atlantic, and ended up in a shed in Ohio. The Wright Brothers didn’t invent precision manufacturing. They inherited it, absorbed it, and applied it to a problem nobody else had solved because nobody else had brought those particular hands to that particular problem.

The chain drive was the hinge. Before it, the bicycle’s design was locked — bigger wheel for more speed, higher and higher off the ground, until the machine teetered at the edge of what a human could survive. The chain drive broke the constraint. It decoupled the pedals from the wheel, let the gearing do what only size had done before, brought the rider back to earth. What had been a machine for athletes became a machine for everyone. What had been the ordinary became, almost overnight, something new.

We are waiting for the chain drive.

Not waiting passively — it is being built right now, in a hundred places at once, by people who mostly don’t know they’re building it. It might be the interface that finally makes AI genuinely accessible to people who can’t do the running vault. It might be the memory architecture that lets a model carry context the way a human carries context, not in a window but in something more like experience. It might be something nobody has named yet, something that will seem obvious afterward, the way all elegant solutions seem obvious after the fact.

What it will not be is the product of people who stayed away from the bicycle shop.

The analyst in Columbus closes her laptop at midnight. The spreadsheet is still not right. She has learned three things about how the model handles date formatting, two things about how it interprets ambiguous column headers, and one thing about her own assumptions that she didn’t know she was making. Tomorrow she will try again. She will get closer. At some point — not tomorrow, maybe not this year — she will get it right, and the thing she learned in the gap will be available to her for the next problem, and the one after that, and she will carry it forward without knowing she’s carrying it, the way craft always travels, in hands that have done the work.

She doesn’t know what she’s riding toward.

That’s the ordinary part. That’s always been the ordinary part.

Categories
AI

The Transit Authority

Today SpaceX went public. The valuation target was $1.77 trillion — already the largest IPO in history, surpassing Saudi Aramco — and the market wanted more.

I was curious about the S-1, so I read the TAM section. SpaceX claims a total addressable market of $28.5 trillion. Rockets and Starlink together account for about $2 trillion of that. The rest — $26.5 trillion — is artificial intelligence. Enterprise AI applications alone: $22.7 trillion.

IDC analyst Arnal Dayaratna said the quiet part out loud: “To be crystal clear, its positioning there right now is basically nonexistent.”

That is an honest sentence. It describes most TAM claims in most S-1 filings. The market did not care. The stock was up 25% anyway.

But the $22.7 trillion number is interesting regardless of whether SpaceX captures it. It asks a real question: how large is the enterprise AI opportunity, really? And what does capturing it actually require?

The answer has something to do with transportation.

We do not travel the same way for every trip.

Walk to the coffee shop. Take a scooter to the office. Ride share to the airport. Commute by train. Drive your own car on weekends. Fly when you need to get somewhere fast and far.

Each mode has a different cost structure, a different latency, a different quality profile. Nobody takes a plane to buy milk. Nobody walks to a meeting in another city. We allocate the mode to the trip, instinctively, without much thought. The routing decision is invisible.

AI inference is arriving at exactly this moment. Until recently, there was one mode: you called the big frontier model. GPT-5.5. Claude Fable. Gemini 3 Pro. You paid the tolls, you waited, and you got what you needed. It was like renting a plane for every trip. Expensive, but simple. There was nothing else on the road.

That is no longer true.

The walk tier is a model running on your phone or laptop — no network, no cost, no data leaving the device. Google’s Gemma 4 and Microsoft’s Phi-4 now handle classification, autocomplete, document summarization. You do not even notice you are using AI.

The bike tier is a small model running on your own hardware — a workstation, a private server. Fast, cheap, data stays on-prem. These models can handle tasks that required GPT-4-class APIs eighteen months ago.

The rideshare tier is cheap cloud inference. You are not driving, not owning, but you get there quickly and cheaply. What cost $22,500 a month in 2025 runs for $405 today. That is not a gradual erosion. That is a structural break.

The car tier is dedicated hosted compute — reserved capacity, predictable performance, always available.

Frontier models are the airplane. Dense reasoning, long-context synthesis, genuinely hard problems. You use them when you need to get somewhere fast and far. You do not use them to classify customer support tickets.

Here is the problem nobody had two years ago.

Picture the IT director at a mid-size insurance company. She deployed a frontier model API last year. Smart decision at the time — one vendor, one contract, everything works. Now she’s gotten the quarterly invoice and done the math. Roughly 80% of the queries hitting that API are things like: extract the date from this document, categorize this claim, summarize this email thread. Tasks a much cheaper model handles just as well. She has been flying everyone to a meeting across town.

She is not alone. Most organizations that built on frontier APIs in 2023 and 2024 are now discovering they over-provisioned for the average query and under-thought the distribution. The expensive mode works. That’s the trap. You don’t look for alternatives when the thing you’re doing works.

The routing layer is where this resolves. A routing layer is need that sits between the application and the model tier and asks, for each incoming query: what does this actually require? Simple queries go to the cheap tier. Hard queries escalate to frontier.

Route 90% of requests to the cheap tier, 10% to frontier. You cut costs by 86%. The quality loss on the 90% is negligible, because most production queries are not frontier-hard. Most trips, you walk.

Back to the $22.7 trillion.

The number is real in the sense that enterprise software currently costs a lot. The global market — CRM, ERP, HR systems, supply chain, all of it — runs roughly $700 billion annually. If AI agents eventually do much of the work those systems mediate, and if the value gets priced into the AI layer, you can arithmetic your way toward very large numbers.

But the routing story embeds an uncomfortable question: if inference costs are collapsing, and if smart organizations route most of their traffic to free or near-free edge compute, who actually captures the value?

The model providers need volume. But enterprise routing gives sophisticated buyers a systematic exit from frontier pricing for the bulk of their workload. You call the expensive plane only when you need to cross an ocean.

This is why the routing layer matters more than it looks. The company that becomes the transit authority — the entity that sits between all the modes and makes the dispatch decision — is structurally positioned to matter as much as any individual model provider. The transit authority does not own the planes or the trains. It knows where you are going and picks the right mode. That intelligence, at scale, is a moat.

SpaceX is not that company. IDC is right about that. But the $22.7 trillion figure, even as a promotional artifact of an S-1, is pointing at something real: the opportunity is large enough that the infrastructure for consuming AI efficiently may be as valuable as the AI itself.

The frontier model providers are the airlines. Necessary, impressive, expensive to operate, essential for the long haul. Emerging routing solutions are building the booking platforms — the systems that decide when you actually need a plane, and make sure you are not buying a first-class ticket to go ten blocks.

In transportation, the booking platforms eventually captured enormous value. Expedia, Booking.com, Google Flights. The airlines, which had all the brand and all the infrastructure, found themselves competing for placement in someone else’s interface.

That story may be ahead of us in AI. The models are the planes. Someone else may be Expedia.