✂️ Accory CCT-75A Cable Cutter delivers clean, precise cable cutting with hardened steel blades, ergonomic grip and an energy-saving design for workshop use. 📧 Contact: wcong0706@gmail.com 🌐 Website: #shunboplastic# #cablecutter# #handtools# #cabletools#

FULL INTERVIEW: Brian Potter says robots learned kung fu before they learned to pick things up because motion is easy to train in simulation and grabbing objects isn't. The fix catching on now is putting "Hulk hands" with cameras on humans to record the data. @_brianpotter writes Construction Physics and is a senior infrastructure fellow at @IFP. He joined @theojaffee and @schisofrenia to break down how AI drives robots, why prefab houses still can't beat hand tools, and the rise and fall of the plasma screen: 00:56 the rise and fall of the plasma screen 04:40 how vision-language-action models drive robots 06:33 why data is the hardest part of robotics 07:46 GPT-6 Astra, and whether frontier models make robot models obsolete 09:51 teleoperation vs simulation 10:18 why robots can do kung fu but can't grab things 11:07 UMI, or giving humans Hulk hands 11:56 how to actually learn how things get built 16:47 why prefab houses still can't beat hand tools 18:12 if we can prefab chips, why not houses 19:07 why cheap robots could fix construction

Some jobs shouldn't have a human in them at all! ⚠️ I visited @mnv_humanoids in Stuttgart, who are building humanoids for hazardous industrial and public safety work. Yesterday they came out of stealth with a ~$10M pre-seed round led by @generalcatalyst, with @LongJourneyVC and @CredoVentures as co-leads. The founders know the category well. CEO @SandorFelber helped launch MIT CSAIL humanoid research line and worked on Tesla Optimus. CTO Maurice Rahme spent six years at Boston Dynamics as one of the technical leads on Stretch. Together they went from an idea to a walking humanoid in five months. Yeah, in 5 months!!! 🤯 One of their examples is the render-safe procedure on an unknown drone at an airport. Today that means a human in a bomb suit walking toward an object nobody has identified yet. Minerva says their robot, Roger, completes the procedure much faster than any existing method. For a warehouse you can redesign the building around the robot, and plenty of companies do exactly that. A refinery cannot be redesigned. Neither can an offshore platform or an airport apron. The ladders, stairs, hatches, valve wheels and hand tools are already there, built around the human body, and retrofitting them costs more than the automation. So the robot fits the environment instead. The control architecture splits the work between human and AI. An operator works from a safe distance, potentially from home, while Minerva Intelligence handles balance, locomotion and manipulation assistance underneath. The human keeps the judgement and the domain expertise, but the robot takes the risk. And every session an operator runs generates training data, so the AI can support them more reliably over time and gradually take on more of the task. → Hardware built in Germany, AI and software developed in the Bay Area → Teleoperation now, with autonomy increasing as the data accumulates → First paid pilots start this fall Put the machine in danger, not the person. That's the whole product. What a time to be a robotics guy these days! 😮‍💨 ~~  ♻️ Join the weekly robotics newsletter, and never miss any news →

‼️A FULL LIST OF ITEMS HIT BY NEW 50% TRUMP TARIFFS, NO USMCA SHIELD THIS TIME ‼️ 👇🏼 Alcoholic beverages (beer, wine, spirits) — Chapter 22 (e.g., headings 2203–2208) Essential oils (limited) Wood and wood products — Chapters 44–46 Paper and paperboard products — Chapter 48 Some miscellaneous items (e.g., certain sports equipment) ----- Agricultural & Plant Products — Honey, feathers/down, animal by-products, bulbs, live plants, cut flowers, seeds, hops, vegetable materials, gums, extracts, etc. Dairy, Sugars & Food Preparations — Milk powders, whey, lactose, glucose/fructose syrups, molasses, bakers’ mixes/doughs. Chemicals & Related — Essential oils, perfumes, cosmetics, polyols (mannitol, sorbitol, glycerol), proteins, glues, starches, paints/inks, lubricants. Plastics & Rubber — Sheets, bags, containers, tableware, household articles, gaskets, seals. Hides, Leather & Furs — Raw hides, leather bags/gloves/belts, fur articles. Wood & Paper — Sawn wood, veneer, plywood, moldings, doors, frames, paper, tissue, cartons, labels. Textiles & Apparel — Fabrics, nonwovens, carpets, clothing items. Machinery, Tools & Electronics — Hand tools, turbines, compressors, refrigerating equipment, packing machinery, cameras, monitors, projectors, radar, smartphones/base stations, printed circuits, etc. Other — Salt, cement, candles, toys/games, furniture, lighting, artwork, antiques, boats, etc. ------- Dairy Products (Chapter 04)0402.10.05 — Milk and cream, unsweetened, in powder/granules/flakes, fat content ≤1.5% 0402.10.10 — Milk and cream, unsweetened, in powder/granules/flakes, fat content >1.5% but ≤3% 0402.10.50 — Milk and cream, unsweetened, in powder/granules/flakes, fat content >3% 0402.21.02 — Milk and cream, sweetened, in powder/granules/flakes, fat ≤1.5%, for infant formula or other special uses 0402.21.05 / 0402.21.25 / 0402.21.27 / 0402.21.30 / 0402.21.50 / 0402.21.73 / 0402.21.75 / 0402.21.90 — Sweetened milk and cream in powder/granules (various fat contents and uses) 0402.91.03 / 0402.91.06 — Condensed/evaporated milk and cream, sweetened, various fat levels 0402.99.68 — Other sweetened condensed/evaporated milk and cream nesoi 0404.10.05 / 0404.10.08 / 0404.10.11 / 0404.10.15 / 0404.10.20 / 0404.10.48 / 0404.10.50 / 0404.10.90 — Whey and modified whey (various types, concentrated or with added sugar) 0404.90.10 / 0404.90.28 / 0404.90.30 / 0404.90.50 / 0404.90.70 — Products consisting of natural milk constituents (whey protein concentrates, etc.) Other Products0506.90.00 — Bones and horn-cores, unworked, defatted, simply prepared 1210.20.00 — Hop cones, crushed, ground, powdered or in the form of pellets 1301.90.91 — Natural gums, resins, gum-resins and oleoresins (other than gum arabic, etc.) 1702.11.00 — Lactose and lactose syrup, containing by weight 99% or more lactose 1702.19.00 — Other lactose and lactose syrup 1702.30.28 / 1702.40.28 / 1702.60.28 — Glucose and glucose syrup (various concentrations) 1702.90.35 / 1702.90.58 / 1702.90.68 — Other sugars, sugar syrups, and artificial honey (fructose, maltose, etc.) 1703.10.30 / 1703.10.50 — Molasses from sugar cane 1703.90.30 / 1703.90.50 — Other molasses 1901.20.35 — Mixes and doughs for the preparation of bakers’ wares 2202.91.00 — Non-alcoholic beer 3301.24.00 — Essential oils of peppermint 3501.10.10 / 3501.10.50 — Casein (various grades) 3501.90.20 / 3501.90.60 — Casein glues and other casein derivatives 3502.20.00 — Milk albumin, including concentrates of two or more whey proteins 3504.00.50 — Peptones and other protein substances and derivatives

Earlier organizations will write their handbooks describing their principles to employees via a handbook of sorts. Amazon had reversible and irreversible distinctions, long-termism and document-reading culture; Facebook had "move fast and break things"; Google had "don't be evil". Now all of us are required to do the same for our fleet of agents via the AGENTS markdown file; the only difference is is that majority of folks have never run a company at scale and don't understand that getting these principles right is highly non-trivial as each one of them has tradeoffs to choose from.

At this intimate counter restaurant, excellent British ingredients meet Japanese technique — and handrolls come with a personal style of service.⁠ Tap here for Jay Rayner's latest restaurant review

On benchmarking long-context agentic instruction following. Agent benchmarks mostly reward reaching the answer. This new benchmark measures whether the agent reached it the permitted way, which is the question enterprise deployments care about. If you ship skills files, policy documents, or long system prompts, you have been trusting that they actually bind agent behavior. But how are you measuring all of this? Surge AI built a benchmark to actually check this. HANDBOOK.md places a standard operating procedure of 20 to 124 pages in context and grades whether it governed every action across an extended tool-use horizon. 65 tasks, five domains, ten fictional companies. Each task runs in a self-contained company environment with a file workspace plus mock email, chat, calendar, issue-tracking, and commerce services exposed over MCP. Every task mutates one of ten base handbooks, altering the specific rules and thresholds that grading turns on, so memorization does not help. Grading is fully deterministic and two-sided. 824 programmatic criteria check that required actions occurred and that prohibited actions did not. Paper: Learn to build effective AI agents in our academy: