Factory Founders Mode: the Team Productivity Model

This week: Architecture for manufacturing systems (paid), team productivity model, Texas REEs, design innovation, history of PI, xLSTM Scaling Laws, Blue Jay and Eluna, 5 Axis 3D Printer, Digital Chem

Oct 26, 2025

∙ Paid

Shop Talk

Capturing this week’s zeitgeist

For industrial technology founders, the recent AWS outage is a critical lesson in business continuity. Your customers in manufacturing, logistics, and robotics are not just buying software; they are integrating your platform into their core operational workflow. When your service goes down, it’s not an inconvenience—it’s a production-halting, supply-chain-breaking event that carries immediate and significant financial consequences for them.

This is precisely why establishing trust with early customers hinges on your ability to guarantee their operational continuity. These initial partners are taking a calculated risk on your new technology or service. The most powerful way to validate their trust is to demonstrate that you are a resilient partner, not just a feature provider. This means architecting your solution to protect their business, even when your own vendors fail. By proactively designing for resilience—through robust offline functionality, rapid failover, or graceful degradation—you are not just selling a product; you are selling reliability. Solving your customer’s business continuity problem is the single most effective way to prove you are a foundational partner, converting that initial leap of faith into a long-term, scalable relationship.

Two recent examples demonstrating founder mode in industrial value chains:

The machine shop founder putting off hiring a CFO because he prefers to run the company on what he calls “Jim math”, making decisions based on what “feels right” for the customer and the product, rather than what is dictated by a spreadsheet /YouTube/
Our freight broker failed us and our only options were to drive across the country or deliver 3 days late, NOX METALS chose driving across the country to get an order delivered on time for one of their most important customers /Zane Hengsperger on X/

📅 Next up: ADIPEC 2025, November 3-6, Abu Dhabi, UAE where dialogue becomes delivery, where solutions are showcased, decisions are made and collaborations are catalysed to unlock long-term value and drive system-wide transformation.

Accelerating the Future of Energy with Physics-Informed AI at ADIPEC /Geminus/

Assembly Line

This week’s Industry 5.0 breakthroughs and frontier technologies of the built world.

🏭👨‍🏭 The Factory Where Employees Are Their Own Bosses

What if a factory could run with zero managers—while producing 23 truckloads of product every single day? Welcome to the Team Productivity Model, a system that’s changing the way American manufacturing works. In this video, we take you inside our shop where welders, fabricators, and builders run the floor, make process improvements, and double wages—all without supervisors. By giving employees ownership of their work, we’ve cut unit build times in half, improved quality beyond overseas competitors, and created a turnover rate of only 2% (vs. the industry’s 30%).

⚒️🪙 A New Generation of Industries Emerges in Texas From Federal Push for Mining Revival

The primary minerals being targeted are lithium, essential for batteries, and various rare earth elements (REEs)—such as samarium, dysprosium, neodymium, and terbium—which are vital for high-tech magnets in military hardware, semiconductors, and renewable energy. The extraction and processing technologies represent a significant shift, highlighted by Direct Lithium Extraction (DLE), where companies pump hot, metal-rich brine from formations (like the Smackover) and use advanced adsorbents and membranes for rapid separation instead of slow evaporation. This is complemented by waste recovery techniques, such as “flash heating” to reclaim minerals from electronic waste and methods to process industrial waste like coal ash and oilfield wastewater. Alongside these innovations, traditional hardrock mining (crushing rock and using acid) and the construction of new advanced refineries are being pursued to process these raw materials into high-purity, battery-grade products domestically within the US.

✏️ Why Simple Everyday Objects Are Impossible to Make

This video uncovers the insane complexity behind the “simple” objects in our lives. Centuries of innovation, billions of dollars in development, and entire global supply chains built the things we take for granted every day.

Covestro wins prestigious Good Design Award for pioneering sustainable design innovation /Covestro/ built on a technical foundation of polycarbonate’s exceptional recyclability.
Clever plastic-free water bottle folds flat when you’re done drinking /New Atlas/

📊 True North with Richard Beeson (CTO OSIsoft)

In a candid conversation on the True North podcast, former OSIsoft CTO Richard Beeson discusses the company’s 32-year history developing the PI System, the “nervous system of industry.” The interview serves as a tribute to founder Dr. Patrick Kennedy, covering key anecdotes about the company’s culture, pivotal product decisions, and Kennedy’s engineering-first ethos.

Discuss on LinkedIn

Predictive Maintenance using AI with Soralink’s Yun Yao /Spotify/ with a solution including smart AI sensors, LTE connectivity, and AI prediction engine.
From Data to Decisions in Days: Litmus Edge on NVIDIA DGX Spark is Accelerating Industrial AI /Litmus/

🧠 xLSTM Scaling Laws: Competitive Performance with Linear Time-Complexity

Scaling laws play a central role in the success of Large Language Models (LLMs), enabling the prediction of model performance relative to compute budgets prior to training. While Transformers have been the dominant architecture, recent alternatives such as xLSTM offer linear complexity with respect to context length while remaining competitive in the billion-parameter regime. We conduct a comparative investigation on the scaling behavior of Transformers and xLSTM along the following lines, providing insights to guide future model design and deployment. First, we study the scaling behavior for xLSTM in compute-optimal and over-training regimes using both IsoFLOP and parametric fit approaches on a wide range of model sizes (80M-7B) and number of training tokens (2B-2T). Second, we examine the dependence of optimal model sizes on context length, a pivotal aspect that was largely ignored in previous work. Finally, we analyze inference-time scaling characteristics. Our findings reveal that in typical LLM training and inference scenarios, xLSTM scales favorably compared to Transformers. Importantly, xLSTM’s advantage widens as training and inference contexts grow.

New Product Introduction

Highlighting new and innovative facilities, processes, products, and services

🚗 Chinese Flying Car Demo in Dubai | Aridge X3-F Takes Flight

🦾 Introducing Blue Jay and Project Eluna, Amazon’s latest robotics and AI technology for its operations

✍️ Author: Tyler Greenawalt

Blue Jay and Project Eluna build on recent advances like Vulcan and DeepFleet, extending Amazon’s approach to physical AI—technology that learns from contact, coordinates at scale, and supports people in the real world. Unlike previous systems (Cardinal, Sparrow, Robin), Blue Jay was developed in just over a year, thanks to heavy use of digital twin simulations.

Blue Jay is an extra set of hands that helps employees with tasks that involve reaching and lifting. It’s a next-generation robotics system that coordinates multiple robotic arms to perform many tasks at once, collapsing what used to be three separate robotic stations into one streamlined workspace that can pick, stow, and consolidate in a single place. The result: more support for front-line employees, while creating greater efficiency in less physical space.

Operations managers constantly monitor dozens of dashboards while responding to technology breakdowns, reallocating resources, and making rapid-fire decisions. Project Eluna acts like an extra teammate, helping reduce that cognitive load. Project Eluna is an agentic AI system—designed to act with a degree of autonomy, reasoning through complex operational situations and recommending actions to operators. It pulls in historical and real-time data across a building to anticipate bottlenecks and keep operations running smoothly.

🖨️3️⃣ Generative Machine Company Makes 5 Axis Desktop 3D Printing a Reality

✍️ Author: Joris Peels

Amidst many copycats, the Generative Machine Company is doing something truly innovative. This British startup is bringing to market 5 Axis Desktop Material Extrusion systems with help from software toolchain firm Ai Build. And just like your favorite cat pictures, they have been made with the help of AI.

Generative Machine makes some of the most beautiful 3D printers I’ve ever seen. Many of the components are made using Powder Bed Fusion in aluminum, so I’m thinking that the printer will be pricey. These components were designed using generative design, specifically using Fusion’s Generative Design tools. That AI bit enabled Ric Real to make these components lightweight and specifically designed to speed up the printer’s printing. But they also mean that the body and motion stages consist of four main parts, simplifying assembly and making this a solid, rigid printer. The printer is powered by Duet3D, and the team is working on closed-loop control for the machine and a photogrammetry capture module that can 3D scan objects for closed-loop control or to print on existing parts.

Business Transactions

This week’s top funding events, acquisitions, and partnerships across industrial value chains.

🇺🇸🔋 Redwood Accelerates Energy Storage, Announces $350 Million Series E Funding

Redwood has closed a $350 million Series E funding round, led by Eclipse with participation from new strategic investors including NVentures (NVIDIA’s venture capital arm).

This oversubscribed round reflects the excitement and rapid growth across Redwood’s two core, integrated business lines:

Critical Materials: Cobalt, Nickel, Copper, Lithium and Cathode Active Material
Grid Energy Storage: very large scale, lowest cost, best integration

By combining deep materials and manufacturing expertise with advanced power electronics and software, Redwood is creating a new generation of U.S.-made energy storage systems—scalable, low-cost, and designed to power data centers, industry, and the grid—reducing reliance on imported LFP batteries.

🏴󠁧󠁢󠁳󠁣󠁴󠁿⚗️ Chemify Raises More Than $50 Million in Oversubscribed Series B to Drive Global Expansion of Digital Chemistry and Discovery

Chemify, the deep-tech pioneer fusing chemistry, robotics, computation, and Chemputation—Chemify’s purpose-built AI to digitize molecule creation—has raised over $50 million in Series B funding to help accelerate its mission of becoming the global leader in digital chemistry, molecular design and manufacturing. The oversubscribed round was co-led by Wing Venture Capital and Insight Partners, with participation from a syndicate of world-class investors including 8VC and existing backers Triatomic Capital, Blueyard, Rockspring and Eos.

Earlier this year, Chemify launched its first Chemifarm, a fully automated chemistry facility built to be the world’s most advanced laboratory for molecular design and synthesis. At its core, the Chemifarm integrates Chemify’s Chemputation platform, an extensible programming language, advanced robotics, and the world’s largest curated library of validated reactions. The platform opens new frontiers in the rapid, data-driven creation of molecules once deemed inaccessible. The new Series B financing allows Chemify to expand this blueprint globally, building a network of digital chemistry hubs and delivering on-demand molecule design and manufacturing to pharmaceutical, biotech and materials-science partners everywhere.

🏴󠁧󠁢󠁥󠁮󠁧󠁿🪙 Milvus Advanced Raises Seed Round to Scale Industrial-Scale Substitution of Critical Elements With Next-Generation Nanoalloy Technologies

Milvus Advanced, a start-up creating rare metal alternatives in the lab through modern alchemy, announced the close of its $6.9 million seed funding round. The investment, led by Hoxton Ventures, includes participation from LQD Ventures, Übermorgen, Tuesday Capital, Mark Leslie Enterprises, van Den Bosch Dynasty Fund, Bluebirds, Md One, EQT Foundation, and returning investor Lowercarbon Capital.

Milvus’s proprietary materials have been independently validated for their superior performance and durability. With this new funding, the company will accelerate its mission to recreate Earth’s rarest metals from abundant elements and scale commercialization across clean energy, transport, electronics, and chemical manufacturing.