Skip to content
#Deployment / engineering
November 10Tuesday
  1. Recode China AI38

    Ulanqab, a Remote Chinese City, Becomes an AI Boomtown and a Magnet for Young Workers

    Ulanqab, a once little-known city in Inner Mongolia, has become an emerging AI computing hub as data centers spread across its grasslands, drawing young workers from across China. Kuaishou began operating its first self-built data center there in 2023, and DeepSeek advertised senior operations and maintenance roles paying up to 30,000 yuan ($4,200) a month this April. The labor market is now saturating, with employers raising hiring standards and some considering only bachelor's degree holders.

  2. Google Blog · AI58

    Google Cloud launches Gemini for healthcare to automate clinical and administrative workflows

    Google Cloud announced Gemini for healthcare, an AI solution that unifies fragmented clinical, operational and administrative systems by connecting to electronic health records, billing platforms, clinical guidelines and Medicare datasets. The platform reasons across that context to orchestrate multi-step workflows such as medical records synthesis and prior authorization review within defined guardrails and with human oversight. It is initially available in preview with launch partners including HCA Healthcare, Highmark Health and Northwell Health, targeting the roughly 28 hours a week clinicians lose to paperwork.

  3. LangChain Blog59

    How Stripe built Kai on Deep Agents in one week

    Stripe built Kai, its company-wide Knowledge AI Platform, on LangChain's open-source Deep Agents harness, with one engineer shipping the initial version in a single week. The stack layers a Stripe-specific harness and configuration layer over Deep Agents, using filesystem middleware backed by S3, sandbox middleware for code execution, summarization middleware for long sessions, and skill-based dynamic tool loading across 500+ internal MCP tools and 1,000+ skills. Kai grew from 296 to over 5,000 users in roughly four weeks and is now used weekly by 83% of Stripe across more than 60k sessions.

  4. Hugging Face Blog60

    Hugging Face explains decision models and their support in Transformers and llama.cpp

    Hugging Face published an explainer on decision models, which take a state plus typed questions and return probabilities over predefined answers instead of generating text. The post details the choice, score and noul primitives, the confidence formula, and new support through Transformers' DecisionPipeline and AutoModelForDecision, a shared /v1/systemone endpoint in Transformers and llama.cpp, a Hub tag for discovery, and fine-tuning with the Trainer API. It notes that the same endpoint also works with compatible autoregressive LLMs by scoring answer logits in a batched forward pass.

    Why it matters: Hugging Face explains how decision models turn classification into a typed, non-generative interface and shows how to serve and fine-tune them with Transformers and llama.cpp.

  5. OpenRouter Announcements72

    OpenRouter benchmarks 15 vision models and Mistral OCR on scanned documents

    OpenRouter tested 15 vision models, Mistral OCR and Tesseract on eight self-made degraded scans, and Claude Opus 5.5 was most accurate with 262 of 264 key values correct at $16.42 per 1,000 pages, while Mistral OCR was best value at 258 of 264 for $2.02. On a faded receipt, GPT-6 Luna and GPT-6.1 Sol invented card digits, a server name and wifi passwords, errors that character error rates do not reveal. The post includes code to transcribe scans, turn reasoning off, cross-check two models from different labs and run Mistral OCR through the file-parser plugin.

    Why it matters: A hands-on benchmark of 15 vision models plus Mistral OCR and Tesseract on eight degraded scans, with runnable code for cross-checking two models to catch fabricated values.

  6. Hugging Face Blog20

    Hugging Face Hub now supports billing through Google Cloud accounts

    Hugging Face organizations can now link to a Google Cloud account so Hub usage appears on the existing Google Cloud invoice instead of a credit card. The setup runs through the Google Cloud Marketplace and covers Team or Enterprise subscriptions, pay-as-you-go compute such as Jobs, Inference Endpoints, Inference Providers and GPU Spaces, plus additional storage at public Hugging Face pricing. Subscriptions are reviewed by Hugging Face before marketplace billing starts, and organizations with a card on file switch at the start of their next billing period.

  7. Google AI · DEV42

    Claude Fable 5.2 Now Available on Google Cloud

    Claude Fable 5.2 is now available on Google Cloud, targeting long-running agents for demanding technical work. The model is positioned for agentic coding across multiple repositories and monorepos, workflows spanning hours to days of unattended work with subagent coordination, and scientific and data pipelines including physics computations, formal proofs, and data analysis.

  8. MIT Technology Review · AI40

    Why the data center backlash is really about electricity

    A visit to Virginia's Data Center Alley suggests the backlash against AI data centers is driven more by electricity demand than by the culture war Big Tech describes. A typical facility there draws 54 megawatts, while the smallest sites attractive for AI training need 300 megawatts and ideal ones about 6,000—roughly Chicago's draw. Companies are turning to onsite gas turbines and diesel generators, and Virginia has imposed new rules requiring data centers to pay fully for new transmission lines.

  9. LangChain Blog50

    LangSmith BYOC on AWS is generally available

    LangSmith Bring Your Own Cloud (BYOC) is now generally available on AWS, giving enterprises a managed LangSmith deployment that runs in their own AWS account and VPC. Sensitive data such as traces, datasets, experiments, prompts, agent deployments and sandbox data stays in the customer's AWS environment, while LangChain handles monitoring, upgrades, scaling and cluster lifecycle management. BYOC is available for Enterprise customers across 15 AWS regions in the US, EU and APAC.

  10. Meta Engineering Blog40

    Meta extends LLDB with vendor-neutral accelerator plugin architecture for CPU-GPU debugging

    Meta has built a vendor-neutral accelerator plugin architecture in lldb-server that lets LLDB debug applications spanning CPUs and GPUs in a single session. The design represents the CPU and GPU as separate LLDB targets, each with its own architecture, threads, registers and address space, while a plugin translates vendor device debugging APIs into LLDB concepts. GPU support activates only when the app uses a supported GPU runtime, so CPU-only debugging keeps the existing native path without added overhead.

  11. Mastra Blog25

    Mastra Launches Mastra FDE Forward-Deployed Engineering Program

    Mastra launched Mastra FDE, a forward-deployed engineering program open to any team building agents, embedding Mastra engineers in customer teams from architecture through production. Engagement ranges from 6 hours per week in a consultative capacity to 20-40 hours per week actively building, backed by a shared Slack channel with Mastra's founders and engineers plus Mastra Enterprise. Mastra says the program costs less than one senior AI hire.

  12. Claude Blog34

    How Curative built a national provider network with agents on Claude

    Curative built a hierarchy of agents on Claude to sign provider contracts about 50 times faster than its team did in 2025, reaching 23,000 contracts signed with its agent Gwen and 2,000 to 2,400 new contracts a week. Gwen handles one provider deal per run, mostly on Claude Opus, while a Sonnet rate reviewer clears about two-thirds of its above-template rate requests and a Haiku judge reads every message before it is sent. Each contract takes about 20 million tokens, roughly a tenth of the original cost, and Curative's contracting team has grown from 45 to about 55 people, each about 2.5 times as productive.

  13. Mastra Blog25

    Mastra Platform Launches $25 Starter Plan With Dedicated Server

    Mastra Platform introduced a $25 Starter plan that raises monthly allowances for compute, data, and observability events and includes one Small dedicated server with its compute covered. Allowances are now higher on every plan, and the free tier is renamed Free, where a serverless agent stays free to run. Projects run serverless or on Small, Medium, or Large servers, and Mastra Connect adds a connection allowance per plan.

November 9Monday
  1. MIT Technology Review · AI12

    Lubrizol's AI-Driven R&D Strategy in Specialty Chemicals

    Lubrizol is building a digital foundation that connects business processes, modern data platforms, and cloud infrastructure to speed up R&D for specialty chemicals. IT vice president Girish Rao says the aim is to give scientists better tools, not replace scientific judgment, with "fusion teams" of scientists, data scientists, and IT staff starting from small pilots to build trust. Cloud access also supports sustainability work such as modeling starch and corn-based material alternatives.

  2. Hugging Face Blog40

    HOPE: Higher-Order Pruning of Experts in Mixture-of-Experts Language Models

    Researchers introduce HOPE, a second-order expert pruning method for Mixture-of-Experts language models that accounts for pairwise interactions between co-activated experts instead of scoring each expert independently. At 50% pruning, HOPE achieves the best average rank of any method, with gains of up to +6.1% on agentic coding tasks over the next-best pruning method. The work targets the memory bottleneck in models such as Qwen3.5-122B, where expert MLPs take up 95% of a ~250 GB footprint.

  3. Anthropic · Claude.dev60

    Anthropic guide on choosing among Claude Fable 5.2, Opus 5.5, Sonnet 5.5 and Haiku 5.5

    Anthropic published a guide recommending developers start most coding and knowledge work on Claude Opus 5.5 at its default medium effort level, since it costs half as much per token as Fable 5.2. The guide maps Sonnet 5.5 to well-scoped everyday work, Haiku 5.5 to high-volume classification, routing and subagent tasks, and Fable 5.2 to the hardest coding, finance and science projects or long unattended runs. It also advises adjusting effort level before switching models, and notes that on SWE-bench Pro, Opus 5.5 at low with failures re-run at high passed about 97% of tasks for about $0.17 each, against 95.3% for $0.29 with every task at high.

    Why it matters: Anthropic's own guide maps each Claude model and effort level to concrete task types, with per-task cost and benchmark comparisons for choosing between them.

  4. Zyphra Research78

    Zyphra releases ZONOS3-Mini, an Apache 2.0 zero-shot voice-cloning TTS model that runs on-device

    Zyphra released ZONOS3-Mini, a sparse MoE text-to-speech model with 100M active and 800M total parameters under Apache 2.0, trained on 112 NVIDIA H100s in 2.5 days. It reaches under 570 ms time-to-first-audio in the browser on a Pixel 9 and under 150 ms on an M3 MacBook Air, with runtimes for WebGPU, SGLang, vLLM and GGML. The model adds control over identity, accent, quality, mood and delivery, including word-level steering, and Zyphra reports it ranks first among open models on its new ZTTS2-Eval benchmark.

    Why it matters: Zyphra's compact MoE TTS model pairs zero-shot voice cloning with word-level delivery control while running in real time on phones and laptops.

  5. NVIDIA Blog38

    NVIDIA Announces Planned Investments in Korean Startups Point2, Motif Technologies, RLWRLD and Upstage

    At the MSIT-NVIDIA AI Startup Partnership Forum in Seoul, NVIDIA announced an investment in Korean AI interconnect startup Point2 and plans additional investments in Motif Technologies, RLWRLD and Upstage. The four companies span the full AI stack, from RF-based chip-to-rack interconnects and open-weight foundation models to robotics foundation models and Solar language models. The planned investments build on NVIDIA Inception, which counts more than 1,000 Korea-based members.

  6. NVIDIA Blog36

    NVIDIA Advances Physical AI in Korea Across Semiconductors, Autonomous Driving and Robotics

    NVIDIA is helping advance Korea's physical AI ecosystem, with Samsung, Kakao Mobility and LG Electronics applying accelerated computing, AI and simulation to semiconductor manufacturing, autonomous vehicles and robotics. Samsung's sMOL platform ran selected quantum chemical calculations up to 5x faster with NVIDIA cuEST, and one AI homogenization case cut simulation time from about 18 hours on a 32-core CPU to 41 seconds. Kakao Mobility plans a level 4 autonomous driving system on NVIDIA DRIVE AGX Thor, while LG demonstrated NVIDIA Isaac GR00T-powered manipulation with its CLOiD robot in a real manufacturing environment.

  7. NVIDIA Blog36

    Korea's AI Ecosystem Advances and Secures Agentic AI With NVIDIA

    Hyundai Motor Group is exploring NVIDIA NemoClaw blueprints and the NVIDIA OpenShell runtime, while AIM Intelligence is integrating its Starfort platform with OpenShell to support secure AI agents for employee productivity and enterprise workflows. OpenShell, a key component of the NVIDIA Open Agent Safety Platform, provides sandboxed execution and policy-based controls to govern autonomous agent actions. LG AI Research is also opening API access for K-EXAONE-2.5, developed using NVIDIA Nemotron datasets including Nemotron-Personas-Korea, an open dataset of 7 million synthetic Korean personas.

  8. NVIDIA Blog34

    NVIDIA and South Korea's MSIT Expand Sovereign AI Collaboration from Chips to Skills

    NVIDIA and South Korea's Ministry of Science and ICT are expanding their sovereign AI collaboration, spanning AI semiconductors, data center infrastructure, talent, startups, models, agents and security. The collaboration opens the way for Korean NPU and XPU companies to explore participation in the NVIDIA NVLink Fusion ecosystem, alongside training, certification, hackathons and technical seminars to strengthen GPU utilization. NVIDIA will also work with three selected consortia and four leading Korean foundation model builders under Korea's AI for All initiative, using NVIDIA Dynamo to optimize and scale AI inference.

  9. Figma Blog38

    Six ways to use the Figma agent on the canvas

    Figma outlined six ways to use its design agent directly on the canvas, from annotating designs with meeting transcripts pulled in via connectors like Granola, Notion, or Google Drive to summarizing comment feedback and applying agreed changes. The agent can also build custom shaders, generate several layout variations in one prompt using a team's library, turn static designs into interactive code layers, and add motion effects.

  10. Google Developers Blog60

    Google Cloud MaxText adds context parallelism for long-context training on TPUs

    Google Cloud released long-context training support in MaxText v0.2.4, its open-source JAX training framework, adding four context-parallel strategies on Cloud TPU: All-gather, A2A (DeepSpeed Ulysses), P2P (Ring Attention) and the A2A + P2P hybrid (USP). Benchmarks with Llama 3.1 8B reached 70% MFU on TPU 5p at sequence lengths up to 4 million tokens, and context extended to 14 million tokens on both TPU 5p and TPU 7, with TPU 7 holding 50% MFU below 6 million tokens. The release is available on PyPI, in Docker images and on GitHub main, with sample recipes for TPU 7 and TPU 5p.

    Why it matters: MaxText's four context-parallel strategies on Cloud TPU, with benchmark MFU figures for Llama 3.1 8B, give teams a concrete reference for choosing how to train at multi-million-token sequence lengths.

  11. Google Developers Blog58

    Google AI Studio adds Google Maps Platform app building from a prompt

    Google AI Studio now lets users build apps powered by Google Maps Platform directly from a prompt, with AI Studio handling configuration and API key management and no credit card required. The integration covers Maps, Places, Routes and Environment APIs, and the sandbox allows 100 calls per day per API with quotas resetting every 24 hours. Linking a Google Cloud billing account unlocks production-grade quotas under standard Google Maps Platform pricing.

  12. Anthropic Research71

    Anthropic tests Claude Fable 5.2 on real-world electronics design

    Anthropic reports that Claude Fable 5.2 designed three electronic devices, a speakerphone, a voice-assistant hub and a handheld virtual pet, from minimal prompts, and that Diode Computers built working prototypes from its design files without circuit fixes. On Diode's internal PCB Gym benchmark, Fable 5.2 scored 69% versus 25% for Fable 5, and it reached 83.5% on the public BenchCAD 2.1, nearly double Claude Fable 5.1. The devices remain prototypes that have not been through electrical validation, antenna tuning, thermal testing or certification, and Claude was not connected to test equipment or manufacturing machines.

    Why it matters: Anthropic's own account of Claude Fable 5.2 designing three working electronic devices shows how far models have moved from partial PCB layouts to end-to-end hardware prototyping.

  13. Google Developers Blog62

    Google makes Pathways generally available on Google Cloud for large-model training

    Google announced the General Availability of Pathways, its distributed training system for large models including Gemini, on Google Cloud Platform. Pathways replaces lockstep SPMD execution with a single Python client that declares the whole computation as a dataflow graph, while a distributed runtime places blocks on virtual devices, moves and reshards tensors, and orders work across the cluster. Google says the design lets a failed host be remapped onto healthy chips instead of restarting the job, and that a shared Pathways service can serve several clients from one cluster.

    Why it matters: Google explains how Pathways' single-controller model differs from SPMD and how it handles cross-group tensor transfer, scheduling and host failure recovery.

  14. Preferred Networks News55

    Preferred Networks releases PLaMo 3.0 Fin finance-specialized small language model as open weights

    Preferred Networks released PLaMo 3.0 Fin, a finance-specialized small language model built on the PLaMo 3 NICT pretraining model from joint research with NICT and further trained on Japanese financial data, publishing open weights on Hugging Face under the PLaMo Community License. The model adds step-by-step Reasoning with reasoning traces output in Japanese, plus tool calling for web search, calculation and internal database access, and can run on a single GPU in closed cloud or on-premise environments. PFN says the instruction-following training for Reasoning and tool use was carried out on its own MN-Core 2 AI chip, with code porting completed within a week and training speed comparable to a GPU environment.

  15. Artificial Analysis Articles58

    Artificial Analysis releases Endpoint Accuracy Index v1.1 with tighter long-context intervals and six-model coverage

    Artificial Analysis launched Endpoint Accuracy Index v1.1, which tightens AA-LCR-25 confidence intervals and adds coverage for GLM-5.3 Flash and Muse Spark Open. Across 103 serverless endpoints benchmarked, 24 score statistically below the self-hosted reference, with causes including output token caps, reduced reasoning, tool call formatting errors and request timeouts. Raising AA-LCR-25 to 25 repeats per question narrows its average 95% confidence interval from ±4.7 to ±2.6 points, and the composite index interval from ±5.5 to ±4.3 points across the 45 endpoints scored under both versions.

November 8Sunday
  1. SemiAnalysis60

    SemiAnalysis visits Envision's 800MW Galaxy datacenter campus in Inner Mongolia

    SemiAnalysis toured Envision's Galaxy campus in Ulanqab, Inner Mongolia, planned for up to 800MW of IT capacity, with its first 120MW phase delivered within a year at about a fifth of US costs. The campus runs a live 5MW 800V DC pilot using solid-state transformers that convert 10kV AC straight to 800V DC for racks, and drops the central chiller by sharing one water loop between liquid- and air-cooled servers. Envision supplies over 60% of the first two phases' power from its own 420MW wind farms plus 4-hour battery storage, and the full campus is expected to be completed by the end of 2027.

November 7Saturday
  1. GitHub Blog · AI & ML38

    Beyond code generation: A developer's guide to agentic engineering

    GitHub published a developer's guide to agentic engineering, walking through how Copilot supports the Plan, Build, Verify, and Observe stages of software work. The guide cites GitHub's 1.2 billion commits in one week this October, over 400% more than a year earlier, and a Faros AI study finding pull requests spent 441% longer in review as teams moved from low to high AI adoption. It details features including Copilot in Slack or Teams, Plan Mode, GitHub Agentic Workflows, cloud sandboxes, session continuity, shared memory, Copilot Dreaming, and Copilot Fusion.

November 6Friday
  1. Ai2 · Allen Institute for AI58

    Ai2 releases ToyBlocks, a controlled simulated world for robot learning and evaluation

    Ai2 released ToyBlocks, an open simulation framework built around a robot arm and blocks on a tabletop, as the first of its FoundationWorlds for controlled robot-learning research. ToyBlocks auto-generates tasks and demonstrations via RL and motion planning in Isaac Lab, reaching up to 24 times real-time data generation on a single GPU, and ships with 85,336 long-horizon demonstrations covering nearly 1,600 hours and 144 million frames. In real-world tests on a Franka FR3, ManiFlow's atomic-skill success was 80.0% in simulation and 81.1% on hardware, while the Molmo 2 plus ManiFlow system succeeded in 55.5% of simulated and 50.0% of real episodes.

  2. vLLM Blog58

    vLLM migrates its CUDA build to the LibTorch Stable ABI

    vLLM has migrated its CUDA build to the LibTorch Stable ABI, and as of v0.33.0 it publishes LibTorch ABI Stable CUDA wheels, so a new PyTorch wheel can be imported without rebuilding vLLM. The team moved in-tree sources and seven LibTorch-dependent dependencies, including FlashMLA, DeepGEMM and FlashAttention-2/3, onto stable APIs, using torch-abi-audit in CI to confirm no unstable symbols remain. The post notes tradeoffs such as a limited stable API subset and small host-side overhead, and says CPU and ROCm builds are next.

  3. OpenClaw Blog50

    OpenClaw adds self-updating model catalog so new models work without an app update

    OpenClaw rebuilt how it discovers models so new ones like GPT-6.1 Sol appear without shipping a release, pulling lists from a self-updating catalog at catalog.openclaw.ai (rebuilt every four hours, about 1,000 models), each account's own provider model list, and a handshake with a newer Codex on your PATH. The catalog switches live on a running Gateway, and model lists no longer shrink after provider timeouts or catalog swaps.

  4. vLLM Blog77

    REINFORCE Zero: Kernel Alignment Enables Stable Async RL at Policy Lag 64, Extends to DeepSeek-V4.1 and Kimi K3.1

    NVIDIA Research and Inferact's REINFORCE Zero paper shows that aligning rollout and trainer kernels lets unclipped token-level importance sampling correct the remaining weight lag, training stably for 20,000 updates at policy lag 64 and reaching 2.40x the training-step throughput of synchronous training with aligned kernels. With standard kernels, the median rollout-to-learner KL rises 83.33x across the run and online AIME24 pass@1 falls from 34.27% to 13.96%, while a sequence gate stabilizes training but rejects 24.48% of sequences by the last 1,000 updates. The zero-mismatch setup now covers DeepSeek-V4.1-Flash, Kimi K3.1 on release day, and DeepSeek-V4-Flash and Kimi K3.1 on AMD MI355X, with code in Inferact/zero-mismatch-rl and NVIDIA-NeMo/labs-molt.

    Why it matters: NVIDIA Research and Inferact split the async RL rollout-learner gap into weight lag and kernel mismatch, showing that aligning kernels lets unclipped importance sampling hold stable training at policy lag 64.

  5. vLLM Blog55

    VeRL-Omni v0.3.0 released with async training and broader multimodal post-training

    VeRL-Omni v0.3.0 was released with asynchronous and decoupled RL execution paths, the V1 trainer as the default, and expanded post-training coverage across image, video, audio, and multimodal understanding. The release adds fully async Qwen3-Omni text and audio training, separate-async Omni and diffusion training with a replay buffer, and new recipes including MiniMax H3 with DiffusionNFT and MiniCPM-o 4.5 GSPO. In a MiniCPM-o 4.5 AVQA experiment on 4×H800 80 GB, separate-async 3+1 was about 9% slower per prompt than colocated 4+0 at natural response length, roughly tied at 512 tokens, and about 6.7% faster at 1024 tokens.

  6. Sierra Blog38

    Sierra Accelerators Take Care of the Basics for AI Agents

    Sierra released Accelerators, prebuilt packages for high-volume workflows like order tracking, appointment scheduling, claims intake and ID verification, available now in Agent Studio. Each Accelerator ships with basic integrations and guardrails, and Ghostwriter adapts it to a company's own policies; updates from Sierra carry over while customer changes are tracked separately. Teams will soon be able to build and share their own Accelerators internally.

  7. SemiAnalysis75

    SemiAnalysis benchmarks Google TPUv7 Ironwood on agentic inference against NVIDIA GPUs

    SemiAnalysis released the first long-context, multi-turn agentic inference results on InferenceX/AgentX for Google TPUv7 Ironwood across four models, finding it competitive on performance per dollar or state-of-the-art in several cases. On GLM-5.2 in FP8, TPUv7 delivered 1.2x to 2.7x the tokens per dollar of B200, B300 and MI355X from 60 to 300 tok/s/user, while FP4 GPUs still led on GLM-5.3 because TPUv7 lacks native FP4 support until TPUv8 in Q1 2027. The piece also argues NVIDIA's bureaucracy slows driver and ecosystem contributions compared with AMD and Google.

  8. Google Developers Blog60

    Google study finds stale TODO bug references distract coding agents in 13% of sessions

    ) references. Agents acted on the bug ID in 6% of encounters, and 83% of those actions were unprompted by the developer; 41% of unprompted detours chased already closed tickets while 57% hit still-open bugs and expanded scope. Controlling for developer, codebase complexity, task type, model tier and other factors, unprompted bug detours added 70% more steps, 77% more tokens and 105% more wait time per session.

    Why it matters: Google's large-scale trace analysis shows stale TODO bug references pull coding agents into unprompted detours, and its cleanup agents removed thousands of obsolete references with over 95% reviewer acceptance.

  9. Modal Blog22

    How Snorkel runs realistic agent simulations with Modal

    Snorkel uses Modal's GPU and CPU sandboxes to run millions of agent simulations per month, peaking at tens of thousands of sandboxes an hour, and relies on Modal for inference on the open-weight models it evaluates. Adopting Modal's memory bursting cut Snorkel's total sandbox bill by around 20%, with roughly 10-15% of its sandboxes using bursting today. Modal was the only sandbox provider able to meet Snorkel's GPU sandbox needs at scale.

  10. Microsoft Research58

    Microsoft Research's PDD Kit formally verifies agent-written full-stack web apps

    Microsoft Research introduced PDD Kit, a research prototype that turns a developer's plain-language intent into a Verus formal specification, proves the coding agent's Rust code meets it for every input, and falls back to tests where a proof cannot reach. In a demo inside GitHub Next's Chopin, the team used PDD Kit to verify the login logic of Pet Store, a roughly 5,000-line Rust full-stack app, with the developer reviewing the intent artifact rather than the generated code line by line. PDD Kit only verifies Rust today, and its guarantees rest on six developer-approved trusted pieces, including the bcrypt password check and the two database lookups.