← All weeks

Weekly digest · 2026-W28

July 6, 2026 – July 12, 2026

7 entries · 33 min total · Foundations & Protocols (5) · Models & Frontier (3) · Infrastructure & Economics (2) · Industry Verticals (2) · Enterprise & Strategy (1) · Governance & Safety (1)

Entries

• Day 112 · July 12, 2026Grok 4.5 Under the Hood -- Token Efficiency vs. TrustThe launch that split into two separate stories -- a genuine cost-math breakthrough for agentic work, and an unresolved fight over who controls what the model is allowed to say
• Day 111 · July 11, 2026The Super-App Wars: ChatGPT Work -- vs. Claude CoworkJuly 9 became the first day three frontier labs shipped at once -- and the real fight moved from chatbot to always-on work agent
• Day 110 · July 10, 2026Agentic AI in Automotive & Autonomous VehiclesRobotaxi fleets, in-cabin copilots, and dealership agents put yesterday's world models to work on the road
• Day 109 · July 9, 2026World Models & Simulation-First AgentsLearning by rehearsing inside a model of the world, before touching the real one
• Day 108 · July 8, 2026Self-Improving Agents 2.0Getting better without retraining the base model Day 100 promised it and Day 101 set it up: today is the technique for how an agent improves itself without ever touching model weights.
• Day 107 · July 7, 2026Claude Science & the Lab-in-the-Loop AI-for-Science StackA technology-first teardown of the AI-for-science agent stack -- and the lab-in-the-loop technique that transfers to any domain.
• Day 106 · July 6, 2026100 Days of Agentic AI -- The Milestone RetrospectiveProtocols, memory, identity, evals, economics, twenty-two industry verticals and a frontier that reorganised itself along the way. Today we grade our own calls -- what the series got right and what it under-called -- distil the ten technologies that mattered most, and set the curriculum for the next hundred days.

Takeaways of the week

• Price agentic workloads per completed task, not per token Compare total tokens burned across an entire agentic run, not the sticker price per million tokens -- Grok 4.5's 4.2x efficiency gap on SWE-Bench Pro would be invisible if you only compared list prices. Day 112
• Pair any efficiency win with your own hallucination check before trusting it A rate that jumps from ~25% to ~54% between versions means the previous version's reliability numbers tell you nothing about the current one -- re-test before routing production work to a new release, however good its cost math looks. Day 112
• Separate "chat" from "work" in your own agent's UX, even inside one app OpenAI kept Chat, Work, and Codex as separate tabs inside one merged app rather than blending them into a single mode -- each has a different session length, permission scope, and failure tolerance. If you're building an agent product, resist the urge to make one mode that does everything; users need to know which mode they're in before they grant it access. Day 111
• Design for the 33%, not just the 9% Coding agents get the headlines, but Anthropic's own usage data shows business-process work (reconciling data, assembling reports) is roughly 4x more common than software development in real Cowork sessions. If you're prioritizing what to automate first inside your org, start with the recurring status-report-and-spreadsheet work, not the flashiest coding demo. Day 111
• Separate your driving/safety-critical agent from your conversational agent, architecturally If you're building anything in a vehicle software stack, keep the perception/planning/control loop and the cabin copilot as two genuinely separate systems with different failure budgets. The industry has converged on this split (Nuro Driver vs. Cerence/Afeela-style copilots) because blending them makes both harder to certify and harder to debug when something goes wrong. Day 110
• Gate any new market or deployment on a real-world validation record specific to that market Whether it's a robotaxi entering a new city or an agent entering a new customer segment, borrow Waymo's employee-first pattern: run a closed pilot under real conditions before opening to the public, and don't assume performance transfers automatically from your first market. Day 110
• Rehearse in simulation before you spend on real-world data collection If you're building or training a robot or physical agent, stand up a synthetic-data pipeline on an open world model (Cosmos-class, via Isaac Sim + LeRobot) before collecting expensive real-world footage. Keep a held-out real-world validation set and re-check the sim-to-real gap every time the simulator or policy changes -- don't assume a policy that works in simulation transfers automatically. Day 109
• Match the world-model architecture to the job, not the hype JEPA-style models (AMI Labs) are the efficient, research-stage option for physical reasoning; latent-action generators (Genie 3) are for playable, interactive environments; explicit-3D reconstruction (Marble) is for when geometric accuracy is the deliverable, not photorealism. Picking the wrong one costs you either compute or accuracy you didn't need to spend. Day 109
• Improve the memory and the skills before you fine-tune For a repeated task distribution, wire MemRL-style retrieval (semantic filter -> learned utility re-rank) before reaching for a fine-tune -- you get continuous improvement with no training infra and, crucially, learning that survives swapping to a cheaper model. Add a skill-writer that distils every multi-step task into a reusable, repairable SKILL.md; keep the library compact and high-utility (SkillFlow shows bloated libraries hurt). Day 108
• Treat every learning write as a deploy A memory write, a skill promotion, and a Darwin-Godel archive update all change the agent's future behaviour, so gate them like a release: OTEL -> WORM audit of what was learned, a human on the promotion gate, and a kill switch that freezes the LEARNING loop -- not just the current run. This doubles as EU AI Act evidence (T-25 to Aug 2). Day 108
• Automate one tight loop, not the scientific method Pick a step where the experiment is cheap, fast and reversible -- literature triage, hypothesis enumeration, an analysis pipeline -- close a loop around it, and keep a human on the physical step and the 'is this real?' judgement. That is where Level 2-3 self-driving labs already deliver. Day 107
• Wrap your best domain models as callable skills The BioNeMo pattern: containerise specialist models (Evo 2, Boltz-2, OpenFold3) as NIM microservices, expose via MCP, and let a generalist agent route to them. Specialist capability arrives as a tool, not a bigger brain -- and the pattern ports to any technical domain. Day 107
• Tomorrow -- Day 101: Claude Science and the lab-in-the-loop Tomorrow -- Day 101: Claude Science and the lab-in-the-loop stack: how AI-for-science architectures actually work -- agentic literature synthesis, hypothesis generation, self-driving-lab integration, and what a drug-discovery agent for neglected diseases does under the hood. Day 106

Viral and open-source picks

• Kling AI's photo-to-dance trend turns any picture into a viral dance video Kling AI's Day 112
• ChatGPT Work goes free on every plan -- including Free Day 111
• Waymo goes driverless in four more cities Day 110
• NVIDIA Cosmos 3 -- the open world model every robotics team is already running The runnable Day 109
• MemTensor / MemRL -- self-evolving memory as runnable open source Day 108
• FutureHouse Kosmos + the Robin lab-in-the-loop system Day 107
• Caveman by JuliusBrussee Day 106

Watch list

• Formal Verification Meets LLMs -- Mistral's Leanstral 1.5 proves code correct with mathematical certainty via Lean 4 instead of just testing it, and what that means for how much you can actually trust AI-written software. Day 112
• Grok 4.5 under the hood -- the token-efficiency claim that changes agentic cost math, and the political-bias benchmark debate that broke out within hours of launch. Day 111
• Agentic AI in Manufacturing & Industrial Robotics -- how the same sim-to-real training loop and three-agent stack from cars extends onto the factory floor, from robotic arms to predictive-maintenance agents on the production line. Day 110
• Tomorrow -- Day 104: Agentic AI in Automotive & Autonomous Vehicles -- how robotaxi fleets, in-cabin copilots, and dealership/service agents put today's world models and sim-to-real training to work on the road. Day 109
• Tomorrow -- Day 103: World models & simulation-first agents (AMI Labs, robotics sim-to-real) -- how agents learn by rehearsing inside a learned model of the world before touching the real one. Day 108
• Day 102: Self-improving agents 2.0 -- MemRL in production, skill-library evolution, and the Darwin-Godel line: how an agent improves without retraining the base model. Day 107
• Day 101: Claude Science + the lab-in-the-loop stack -- agentic literature synthesis, hypothesis generation, self-driving-lab integration, and the drug-discovery agent architecture under the hood. Day 106