Michael Rudoy

I then layered in Paleologo/Bouchaud physics to size fat-tail risk, guarded by Wes Gray momentum failsafes and De Prado/Isichenko meta-labeling. Fundamental wisdom, executed with quantitative ruthlessness. (2/2)

I just built the ultimate quantamental machine. It uses an ensemble of 7 fundamental brains (Buffett/Fisher) to find quality, and a Seth Klarman "Dictator" to ruthlessly veto bad valuations. (1/2)

@espn Inherited him as my coach at Princeton. Forget basketball — one of the most difficult (being diplomatic) people I’ve ever come into contact with. Always expected this would happen as a result.

Air Force men's basketball coach Joe Scott has been indefinitely suspended pending an investigation into the treatment of cadet-athletes. spr.ly/6013C2t31

@Polymarket @grok true?

BREAKING: Zohran Mamdani to require all New York elementary school students to learn Arabic numerals.

@stuartawillson The knowledge doesn’t leave at once but hear you

@mrudoy creating all that info Mike!

Alex Gruebele from Tau9 Labs told me that “A quarter of the workforce is about to retire. All their knowledge gets lost unless you capture it." A quarter of the manufacturing workforce is about to retire! And they are the people who know the machines, the failure modes, the workarounds, the clues in the sound or smell or vibration, the tribal fixes that never made it into an SOP. When they walk out 2the door, most of that knowledge goes with them. You can’t hire your way out of it. You can’t train your way out of it. You can’t replace 30 years of intuition with a new dashboard. But! Most of the knowledge already exists somewhere. It lives in: • technician reports • warranty claims • part orders • failure logs • scrap reports • photos and PDFs • ERP exports • the notes no one reads • the emails everyone forgets • the “ask Jim, he always knows” conversations Alex told me: "AI can turn that knowledge into a searchable memory.” AI can finally read all of it. It can summarize it, pattern-match it, spot anomalies, surface root causes, and preserve what used to be tribal. There's a lot of anxiety about AI replacing jobs, and there are definitely jobs that are "high risk," but this isn’t about “AI replacing jobs.” It’s AI capturing and preserving expertise that would otherwise disappear forever. For manufacturers, that is the difference between: [1] Preventing downtime or repeating it [2] Fixing a defect or chasing it [3] Training new hires in weeks instead of years [4] Scaling knowledge instead of losing it This seems like one of the biggest under-the-radar opportunities in industrial AI right now. And it’s happening quietly, one workflow at a time. Go get it.

@stuartawillson CS

Stories like this make me super bullish on custom software. In my Just Curious conversation with Alex Gruebele, PhD and Chandler Gonzales (co-founders of Tau9 Labs), we dig into how they helped an injection-molding manufacturer replace an off-the-shelf quoting tool with a custom AI system built in ~3 months, cutting quoting time by 90%+ and doubling win rates. This isn’t a generative/agentic story. It’s expert developers using AI-assisted coding (aka “vibe coding”) to change the economics of custom software: lower build cost, faster time-to-value. That trade-off used to favor off-the-shelf. But not anymore. And especially not within manufacturing where every factory is unique. As Alex says, “Code is cheap now. The hard part is knowing what to build.” And Chandler adds: “Manufacturers don’t need another SaaS dashboard. They need AI that understands their context—their machines, their materials, their people.” A few more things of note: [1] Context > code. Every factory is unique; messy people, machines, and data. Generic SaaS breaks on impact. Tau9 walks the floor, sits with operators, and maps how parts really move. [2] Follow the five whys. When a client asks for “AI for drawings,” keep pulling the thread until you hit the real problem. The best AI projects start with workflow discovery, not a prompt. [3] Build for speed, not scale. From concept to launch in a few months, with a 7-click path from 3D-file upload to purchase. AI-assisted coding made development 3–5× faster, and speed converted to revenue. [4] On-site beats offshore. Walking the line reveals the unspoken context—the unplugged cable, the sticky workaround—that remote teams never catch. This is part of a bigger shift: small, specialized teams + AI copilots delivering high-ROI custom software faster than any legacy vendor can ship an update. Tomorrow I'll share the step-by-step case study on how Tau9 built the system, and the lessons manufacturers (and investors) can learn from. Enjoy!

@stuartawillson True!

You wouldn’t hire without a reference. You wouldn’t marry without dating. So why bet your AI strategy on the first partner you meet? We built a way to run a free, blinded mini-RFP for your AI project. And then get matched with vetted applied-AI experts trusted by firms from Red Antler to Mayo Clinic. justcurious.io/get-expert-mat…

@aaditsh @stuartawillson

MIT published a 26-page report on AI. Save this one.

@stuartawillson Then don’t

@mrudoy I'm not going to read this, Mike.

They said don't read the comments. But now the whole internet is just the comments.

@stuartawillson @FedEx Why not

@stuartawillson Agree loved it

American Masters on PBS is such a great docuseries. Just watched the Geffen one, Inventing David Geffen, which is an amazing tale. Hustle. Ruthlessness. Great taste. Complex guy. Amazing career. Amazing time in music and in Los Angeles.

@stuartawillson Thx

Here's both a six-step AI deployment framework and an example of how a financial advisor platform used the results to increase advisor adoption from 25% to 80%, save time (full day a week), and more. Back for round two with Jordan Gurrieri, Co-founder and CEO of BlueLabel. In our first conversation, Jordan broke down why most AI pilots fail to scale. This time, he walks through how BlueLabel fixes that by using their six-step SPRINT framework and also shares an example of how SPRINT delivered measurable ROI for a financial advisor platform. First, the SPRINT Framework: A six-step blueprint for turning AI strategy into real results: [S] Strategic Alignment – Start with business goals, not technology. AI should serve clear outcomes. [P] Process Mapping & Prioritization – Map workflows, find the biggest pain points, and rank opportunities. [R] Rapid Experimentation – Build and test lightweight pilots early and often. [I] Integration & Insights – Learn from real use, refine the roadmap, and improve the AI itself. [N] Now–Next–Later Roadmap – Sequence quick wins and long-term transformations to maintain momentum. [T] Transfer & Transformation – Build internal champions to own and scale the work. Then, the Case Study: Scaling a Wealth-Management Workflow A family office offering retirement planning and wealth management services, supported by a network of financial advisors came to BlueLabel with a big gap: their signature “plan path," a multi-step onboarding and planning process, was so manual that only 25% of advisors actually followed it, leaving growth and consistency on the table. Using the SPRINT framework: [1] The team mapped the advisor workflow and identified friction points like duplicate data entry and lost client language. [2] They ran rapid experiments to use AI for parsing reports and rewriting plan documents in the client’s voice. [3] Within weeks, the prototype saved each advisor 30 minutes per plan. [4] BlueLabel scaled it to all advisors through a phased Now–Next–Later roadmap. [5] Senior advisors became internal champions, training new hires and driving full adoption. The results: ✅ Advisor adoption jumped from 25% to 80% ✅ Efficiency improved by ~20% (a full day per week saved) ✅ Close rates improved by 5 points, adding millions in new AUM Watch the video below. thx, Stu

@stuartawillson Silly goose

@stuartawillson Didn’t get there yet

Only real disappointment with the Barry Diller autobiography is that the Interactive Corp / IAC period is largely skimmed through at the end.

@stuartawillson True

What a scam this Hertz PlatePass is. Show me a 'convenience charge' / processing fee that is 3.4x the transaction being processed and I'll show you a company that is long-term ngmi.

@stuartawillson Did it heal?

The downside to having a spring in your step is hitting your forehead on the 2nd-floor overhang as you bounce down the stairs. Dented the overhang and cut my forehead open.

@stuartawillson I love them too

Taking advantage of their underdeveloped taste buds, I give my 4 and 6-year-olds all the Raspberry Spindrifts  –and I'm ok with this.