Hot Trending Topics

Physical Intelligence introduced π₀ (pi-zero), a pioneering general-purpose vision-language-action (VLA) foundation model designed to bring physical robots closer to general intelligence. Built on top of pre-trained vision-language models like PaliGemma and paired with a high-frequency flow-matching action expert running at up to 50 Hz, π₀ enables multi-task robotic manipulation across diverse hardware platforms without task-specific hardcoded scripting. Viral demonstration videos showcased robots powered by π₀ folding laundry straight out of a dryer, clearing cluttered dining tables, sorting miscellaneous trash, and assembling cardboard boxes. The release has captivated the AI and robotics community because it represents a major conceptual transition from rigid, pre-programmed robotic movements to internet-scale semantic reasoning combined with fluid physical reflexes.

Boston Dynamics released groundbreaking footage of its fully electric Atlas humanoid operating autonomously inside Hyundai Motor Group's Metaplant America (RMAC) in Georgia. Moving decisively past the hydraulic parkour prototypes of the previous decade, this commercial-grade Atlas is seen identifying, gripping, and sorting heavy automotive structural parts and roof racks weighing up to 30 kg without any human teleoperation. Utilizing advanced computer vision, end-to-end neural motion policies, and thousands of hours of parallel simulated training in digital twins, Atlas continuously reorients parts on the fly to match vehicle assembly fixtures. The demonstration is trending globally as concrete evidence that enterprise humanoids are entering revenue-generating assembly lines to perform ergonomically punishing factory work.

Tesla made headlines across tech and financial media following updates regarding its Optimus humanoid robotics program. During recent executive briefings, Tesla announced the commissioning of an early mass-production line at its Fremont factory—occupying floorspace previously allocated to legacy vehicle assembly—focusing on the newly refined Optimus Gen 3 hardware. Featuring all-new 22-degree-of-freedom hands with integrated tactile sensors, upgraded electromechanical actuators, and preparation for Tesla's onboard AI5 chip, the robot is currently being deployed into Tesla's own manufacturing facilities through the internal 'Optimus Academy' training curriculum. The announcement generated massive social media engagement and debate regarding Tesla's targeted commercial delivery timeline of late 2027.

Figure AI and the BMW Group announced the successful conclusion of their 11-month Figure 02 pilot at the Spartanburg assembly plant—which logged over 1,250 runtime hours and directly contributed to the assembly of more than 30,000 BMW vehicles—and unveiled the next phase with Figure 03. Driven by the new Helix 02 visual-motor AI system, Figure 03 demonstrates whole-body dynamic coordination, moving from body shop sheet-metal insertion into Hall 52 for high-speed assembly and logistics sequencing. Trending video clips show the robot coordinating footsteps while pulling heavy caster-wheel carts and precisely depositing components into sequencing trolleys, showcasing one of the most rigorous real-world validations of commercial humanoids in automotive history.

1X Technologies, the Norwegian-American robotics firm backed by OpenAI, unveiled NEO Beta, a bipedal humanoid engineered specifically for consumer homes rather than industrial warehouses. Ditching exposed industrial metal and pinch points, NEO features a soft-body polymer lattice exterior, biomimetic tendon-driven actuators, and an ultra-quiet lightweight profile weighing just 30 kg. Viral demonstrations show NEO Beta interacting safely with family members, picking up clothing from floors, folding blankets, clearing kitchen dishes, and conversing using natural voice prompts. The announcement that NEO will enter select home trials with commercial pricing around $20,000 (or $499/month subscription) sparked widespread discussions about consumer home robotics finally becoming a reality.

Unitree Robotics made waves across the global physical AI landscape by scaling up mass production of its Unitree G1 humanoid robot and announcing a 'self-evolving' physical AI framework. Priced aggressively starting around $16,000 for base models and unlocking full 43-DoF dexterity and developer SDKs in EDU tiers, the G1 can sprint at 2 m/s, execute dynamic backflips, and recover instantly from falls. Unitree founder Wang Xingxing revealed that new automated reinforcement learning loops allow the G1 to write its own motor code, run thousands of iterations in physics simulators, and deploy physical behavioral updates with minimal human intervention, making it a favorite platform across global AI research labs.

OpenAI launched 'Operator', an autonomous computer-using agent capable of independently executing complex, multi-step digital tasks directly inside a web browser and operating system. Unlike simple text assistants, Operator interprets screen visuals, moves the virtual mouse cursor, clicks elements, fills out multi-page booking forms, conducts market research, and debugs software workflows autonomously. Tech analysts and developers flooded social media with demonstration videos showing Operator booking complete flight and hotel itineraries, navigating messy corporate dashboards, and ordering specialty groceries online, heralding the rapid dawn of agentic web automation.

Anthropic's release of the 'Computer Use' API for Claude 3.5 Sonnet continues to trend heavily as developers build complete desktop autonomous agents. The groundbreaking model can look at computer screenshots, calculate exact pixel coordinates, generate keystrokes, scroll windows, and interact seamlessly with native desktop software and legacy enterprise applications that lack traditional APIs. Coding enthusiasts and tech YouTubers demonstrated Claude taking over multi-monitor setups to write code in IDEs, execute terminal commands, test user interfaces in real time, and fix bugs autonomously, setting a new benchmark for multimodal desktop agency.

DeepSeek's release of DeepSeek-V3 and the DeepSeek-R1 deep reasoning model triggered massive shockwaves throughout Silicon Valley and financial markets. DeepSeek proved that frontier-class reasoning models—competing toe-to-toe with proprietary giants like OpenAI o1 across advanced mathematics, competitive programming, and STEM benchmarks—could be trained with pure reinforcement learning (RL) and mixture-of-experts (MoE) architectures at a fraction of traditional hardware costs. The open release of R1 model weights and distilled variants ignited an explosive global wave of local AI deployments, academic scrutiny, and discussions on how open-source AI is disrupting high-capital tech monopolies.

At its premier developer keynote, NVIDIA showcased the latest breakthroughs in its robotics and physical AI ecosystem. CEO Jensen Huang captivated audiences by demonstrating tiny 10-inch bipedal 'Microduck' robots learning dynamic locomotion in Isaac Sim via the Newton physics engine alongside the Project GR00T foundation platform and Jetson Thor modules. NVIDIA also unveiled Cosmos, a suite of world foundation models engineered to generate photorealistic physics-accurate training environments for autonomous robots and self-driving vehicles, dramatically speeding up sim-to-real transfer and reducing real-world data collection requirements.

Google DeepMind unveiled its Gemini Robotics and Gemini Robotics-ER (Embodied Reasoning) architectures, bringing the spatial intelligence of multimodal Gemini foundation models directly into physical robotic arms and mobile manipulators. The system allows robots to understand open-world 3D spaces from raw RGB-D camera feeds, infer implicit physical relationships (such as fragile items, centers of gravity, and liquid contents), and execute complex multi-stage tasks commanded in casual conversational English. Videos showing robots preparing breakfast trays and sorting delicate items in cluttered kitchens quickly went viral on YouTube and X.

Alphabet's autonomous driving unit Waymo announced a major commercial milestone as its 6th-generation Waymo One driverless robotaxi service officially surpassed 150,000 paid passenger trips per week across Phoenix, San Francisco, Los Angeles, and Austin. Waymo also launched commercial driverless airport pickups at Phoenix Sky Harbor and began testing highway speeds with fully unoccupied vehicles. Tech commentators and riders posted thousands of on-board ride videos praising the silky-smooth driving behavior and zero-safety-driver reliability compared to human rideshares, solidifying Waymo's lead in commercial autonomy.

Agility Robotics made headlines as its bipedal humanoid robot Digit was expanded into active production environments at agricultural machinery giant AGCO and Amazon fulfillment hubs. Operating out of the company's RoboFab production plant in Oregon, Digit humanoids are assigned to repetitive bulk container tote management, picking up plastic bins from conveyors and stacking them precisely onto automated transport dollies. The deployment is celebrated as one of the first multi-robot bipedal fleets integrated into active warehouse software workflows without requiring floor infrastructure overhaul.

Sanctuary AI revealed major progress in general-purpose manipulation with its Phoenix humanoid powered by Carbon, an embodied AI cognitive system. Phoenix demonstrated human-like hand speed and precision across hundreds of real-world retail and industrial tasks, including operating point-of-sale scanners, bagging groceries, sorting medical supplies, and folding garments. The company's demonstration of autonomous dexterous teleoperation data collection and real-time sim-to-real tactile feedback received praise from robotics engineers for closing the gap between artificial cognition and biological motor control.

Austin-based robotics startup Apptronik showcased the latest milestones of its Apollo humanoid robot trial with luxury automaker Mercedes-Benz. Standing 5 feet 8 inches and capable of lifting 55 pounds, Apollo is deployed along assembly lines to deliver component kits to human manufacturing workers, inspect assembly parts, and handle repetitive logistics. The collaboration highlights how automotive manufacturers are turning to humanoid cobots to address severe industrial labor shortages while ensuring ergonomic safety in high-precision assembly lines.

Meta unveiled Movie Gen, a breakthrough research model capable of generating high-definition 1080p video with synchronized cinematic audio, sound effects, and ambient music from natural language prompts. Concurrently, Meta rolled out real-time multimodal AI updates to its viral Ray-Ban Meta smart glasses, allowing users to look at physical objects, landmarks, or text and receive conversational voice guidance through open-ear speakers. Influencers and tech creators widely shared clips of the smart glasses acting as an instant AI tour guide and translator, highlighting the convergence of wearable AI and everyday physical interaction.

Intuitive Surgical's newly deployed da Vinci 5 robotic-assisted surgical system is trending among medical professionals and roboticists. The next-generation surgical robot incorporates high-resolution 3D endoscopy, over 10,000 times the computing power of its predecessor, and true Force Feedback technology that enables surgeons to feel physical resistance through robotic controls during delicate procedures. AI algorithms running in the background analyze surgical movements in real time to suggest ergonomic optimizations and reduce tissue trauma, marking a giant leap forward in surgical robotics.

Researchers at MIT published breakthrough robotics findings introducing insect-scale aerial micro-robots capable of unmatched agility and acrobatic maneuvers. Powered by soft artificial muscle actuators and an ultra-lightweight neural flight control policy, these centimeter-sized drones can perform multiple continuous somersaults in fractions of a second, hover in turbulent drafts, and recover from direct mid-air collisions. Video footage of the tiny drones darting through obstacle courses captured massive interest online, pointing to future applications in search-and-rescue, hazardous industrial inspection, and micro-pollination.

Engineers at the Technical University of Munich (TUM) unveiled a groundbreaking neuromorphic AI edge processor designed specifically for autonomous robots. By mimicking the spiking neural networks of biological brains, the processor executes computer vision, real-time spatial mapping, and motor control using up to 90% less power than conventional GPU clusters. This allows mobile robots and drones to navigate completely offline in GPS-denied and disconnected environments without transmitting sensitive sensor data to the cloud, resolving critical latency and battery constraints in physical AI.

Disney Research and Walt Disney Imagineering released viral footage demonstrating their BD-X bipedal robotic droids walking expressively and interacting with park guests. Developed using GPU-accelerated reinforcement learning inside simulated physics engines, the pint-sized robots can convey emotional states—such as curiosity, joy, or apprehension—entirely through body language, stride cadence, and head tilt while maintaining rock-solid dynamic balance over uneven terrain. Robotics and animation enthusiasts praised the system as the pinnacle of merging artistic character design with state-of-the-art robotic locomotion.

Elon Musk's brain-computer interface (BCI) company Neuralink released comprehensive video updates documenting patient trials with the 'Telepathy' neural implant. In demonstrations that stunned neuroscientists and the public alike, human trial participants paralyzed by spinal cord injuries controlled digital computer cursors, played competitive video games, and maneuvered multi-axis robotic arms purely through thought. The seamless decoding of neural motor intentions using high-density microelectrodes represents a monumental breakthrough for assistive robotics and medical neurotechnology.

LimX Dynamics introduced TRON 1, a groundbreaking bipedal robot featuring a point-foot, sole-foot, and wheeled-foot modular design. Engineered as an open research and development platform for embodied AI, TRON 1 effortlessly navigates staircases, slippery ramps, and gravel trails using proprioceptive reinforcement learning controllers. Viral clips showing the wheeled-biped version performing seamless 360-degree spins, stair climbing, and recovery after being kicked highlighted how modular hardware is lowering barriers for robotics locomotion researchers worldwide.

Chinese EV maker XPENG unveiled its flagship humanoid robot, 'Iron', at its annual tech day and confirmed the robot is already integrated into active EV assembly lines. Standing 178 cm tall and equipped with XPENG's proprietary Turing AI chip boasting 40-core neural processing and end-to-end vision-language models, Iron features 62 degrees of freedom and tactile robotic hands capable of assembling small mechanical components. Automotive analysts highlighted that XPENG is leveraging its self-driving vehicle AI models and manufacturing infrastructure to build a competitive edge in enterprise humanoids.

Fourier Intelligence debuted GR-2, the next generation of its general-purpose humanoid robot. Featuring an elegant ergonomic chassis, 53 degrees of freedom, and biomimetic high-torque actuators, GR-2 is tailored for physical therapy, medical rehabilitation, and assistive nursing. The robot's hands possess 12 degrees of freedom with integrated force sensors, enabling it to assist patients standing up, carry medical equipment, and manipulate fragile glassware. The release reflects how service humanoids are rapidly evolving beyond industrial factories into healthcare and elderly assistance.

Kepler Robotics announced the commercial release of the Forerunner K2 humanoid robot, targeted at hazardous industrial environments, mining sites, and automated petrochemical facilities. Powered by the 'Nebula' AI embodied platform and equipped with an array of LiDAR and thermal sensors, Forerunner K2 can lift 25 kg in each arm, navigate rugged terrain, and perform autonomous routine safety inspections around high-voltage machinery. The demonstration of heavy-duty industrial humanoids operating in hazardous environments without endangering human lives gained substantial industry traction.

Mentee Robotics, co-founded by computer vision pioneer Amnon Shashua, published new demonstration footage of MenteeBot executing complex household and warehouse tasks guided entirely by vision-based AI. Rather than using expensive LiDAR sensors, MenteeBot relies on standard RGB cameras, computer vision transformers, and end-to-end neural motion policies to map environments, understand spoken user commands, and retrieve items from multi-story shelving units. Commentators celebrated Mentee's cost-effective vision-first philosophy as a realistic blueprint for commercial robotic scalability.

Carnegie Mellon spin-out Skild AI surged in popularity across tech blogs following significant venture funding and technical reveals of 'Skild Brain', a universal foundation model for physical robotics. Skild Brain is designed to serve as a shared operating intelligence across humanoids, quadrupeds, and industrial robot arms alike. Demonstrations showed robots using Skild Brain adapting instantly to novel manipulation tasks, navigating obstacle-laden staircases, and manipulating deformable items like cloth and cables without needing task-specific training data.

Hugging Face's LeRobot initiative continues to explode across the open-source developer ecosystem, empowering roboticists to build, train, and share physical AI models using standard consumer hardware and affordable robot arms like SO-100 and Koch. LeRobot provides standardized datasets, pre-trained imitation learning checkpoints, and seamless integrations with PyTorch and Hugging Face Hub. Community members shared thousands of videos showing sub-$200 robot arms learning to pick up berries, sort pens, and pour drinks using imitation learning, democratizing physical AI for makers worldwide.

K-Scale Labs attracted immense interest from the DIY and academic robotics communities after releasing complete, open-source mechanical CAD files, electrical schematics, and reinforcement learning software for its humanoid robot, Stompy. Designed to be 3D printed and assembled using off-the-shelf brushless motors and low-cost actuators, Stompy lowers the cost barrier of developing bipedal humanoids from tens of thousands of dollars to just a few thousand. Tech communities and GitHub repositories exploded with forks and collaborative builds aiming to create the Linux of humanoid robotics.

Robotics titan ABB and technology investment powerhouse SoftBank announced a multi-billion-dollar global strategic alliance to integrate multimodal generative AI and robotic foundation models into industrial automation. The joint initiative aims to deploy vision-language-action policies across ABB's vast installed base of industrial cobots and factory arms, enabling operators on factory floors to instruct assembly robots in plain language and have the robots generate, simulate, and execute new assembly paths in seconds.

Swiss robotics firm ANYbotics showcased successful autonomous commercial deployments of its ANYmal quadruped inspection robot across offshore energy platforms and chemical plants. Equipped with explosion-proof IP67 hardware, acoustic leak detection microphones, thermal cameras, and onboard SLAM, ANYmal traverses steep metal stairs, grated catwalks, and tight corridors in complete darkness without human oversight. The footage of quadruped robots tirelessly inspecting high-risk industrial sites sparked viral discussions on replacing dangerous offshore human shifts with autonomous physical AI.

Boston Dynamics shared significant operational updates regarding Stretch, its specialized box-unloading mobile robot currently deployed in major logistics facilities like DHL. Stretch now utilizes generative vision models and real-time physics heuristics to autonomously unload haphazardly stacked shipping containers containing deformed, shifted, or unlabelled cardboard cartons without jamming. Video of Stretch dynamically grabbing falling boxes mid-air and maintaining speeds of over 700 boxes per hour went viral across supply chain forums.

Amazon announced that its fully autonomous mobile robot Proteus has completed over one billion package movements across North American and European fulfillment networks. Unlike legacy automated guided vehicles restricted to caged zones, Proteus uses advanced 3D spatial AI and perception safety suites to navigate seamlessly alongside human employees on open warehouse floors, while the Sparrow robotic arm handles individual product picking. Amazon's milestone videos triggered lively debate regarding warehouse productivity and the accelerating pace of supply chain automation.

Amazon's purpose-built robotaxi subsidiary Zoox began offering public, uncrewed commercial rides in its iconic bidirectional autonomous vehicle along the Las Vegas Strip and parts of the San Francisco Bay Area. Featuring symmetrical four-wheel steering, facing passenger seats, and no steering wheel or pedals, Zoox vehicles navigate high-density pedestrian zones, complex multi-lane merges, and casino pickup lanes entirely via its custom 360-degree sensor suite and deep neural planners, attracting glowing reviews from first-time riders.

UK-based autonomous mobility startup Wayve captured global headlines with GAIA-1 and LINGO-2, end-to-end generative world models for embodied artificial intelligence. GAIA-1 functions as a world simulator capable of generating realistic future driving video sequences based on natural language prompts and vehicle trajectory inputs, while LINGO-2 allows autonomous vehicles to explain their driving reasoning in real-time conversational English. Tech analysts highlighted Wayve's foundation model approach as a fundamental departure from rigid, rule-based autonomous vehicle stacks.

Reports from Bloomberg and tech insiders revealed that Apple has significantly expanded its confidential robotics engineering team, focusing heavily on a smart home tabletop device that pairs an iPad-like display with a motorized robotic actuator arm. Driven by Apple Intelligence and computer vision, the robotic arm will mimic human head movements during FaceTime video calls, track occupants around the room, and manage home automation hubs. The leaks sparked widespread excitement and speculative video renders about Apple entering the physical AI ecosystem.

Conversational search engine Perplexity AI launched a comprehensive AI shopping experience featuring instant product comparisons, unbiased reviews synthesized from Reddit and expert forums, and one-click 'Buy with Pro' agentic checkout. Users can query Perplexity for niche product recommendations and authorize the AI agent to complete the entire purchasing transaction on merchant websites without leaving the interface. E-commerce analysts noted that Perplexity's agentic commerce feature directly challenges Google's traditional sponsored search ad business model.

The phrase 'vibe coding'—coined by former OpenAI chief scientist Andrej Karpathy—has become the hottest software engineering trend on GitHub, YouTube, and X. Driven by AI-native IDEs like Cursor and Windsurf integrating Claude 3.5 Sonnet and DeepSeek, developers describe building full-stack software applications and complex web apps purely through natural language prompts and multi-file agentic auto-completion without writing syntax manually. Discussions on how AI coding agents are shifting engineering from manual programming to architectural orchestration dominated developer feeds.

Cognition AI announced enterprise upgrades to Devin, its autonomous AI software engineer. Devin can now ingest entire private corporate codebases, set up developer environments, autonomously reproduce bug reports submitted via Jira or GitHub Issues, write unit tests, and submit clean pull requests for human review. Live demonstration streams showcasing Devin migrating complex legacy codebases and resolving production errors with minimal human intervention reignited discussions on the evolving role of software engineers.

Groq generated immense buzz across AI and hardware engineering channels by demonstrating ultra-high-speed inference on its Language Processing Units (LPUs). Achieving token generation speeds exceeding 500 to 1,000 tokens per second on open-weights foundation models, Groq's deterministic architecture is being leveraged by physical robotics developers to eliminate inference lag in vision-language-action policies. This enables robots to process visual sensory data and execute reactive motor adjustments instantaneously, addressing the latency bottleneck in physical AI.

Cerebras Systems unveiled new cloud-scale AI supercomputing benchmarks for its CS-3 Wafer-Scale Engine, the largest commercial computer chip ever built. Packing 4 trillion transistors and 900,000 AI cores onto a single silicon wafer, the CS-3 is being utilized by robotics research consortiums to simulate massive multi-agent physics environments in real time. The ability to simulate millions of virtual robot hours per day without multi-GPU interconnect bottlenecks has accelerated the training of generalized robotic motor policies.

Runway AI trended across digital media and filmmaking communities with the rollout of Act-One and Gen-3 Alpha Video-to-Video. Act-One allows creators to record simple smartphone facial performances and instantly transfer micro-expressions, eye movements, and emotional delivery onto photorealistic or stylized animated characters without motion capture gear. Digital creators flooded YouTube and TikTok with side-by-side comparison reels showcasing how AI video generation has transitioned from erratic morphing to production-grade Hollywood character animation.

Paris-based AI pioneer Mistral AI unveiled a massive revamp of its Le Chat conversational platform, introducing real-time web search with citations, deep document and code analysis, and native vision capabilities powered by Pixtral 12B. With open weights released under flexible licenses, Pixtral processes high-resolution images, architectural diagrams, and technical schematics with frontier-tier accuracy. European tech leaders and global developers celebrated the release as a major boost for sovereign, open-source AI alternatives to American tech giants.

Microsoft introduced autonomous agent creation capabilities inside Copilot Studio, allowing businesses to build independent AI agents that operate proactively in the background rather than merely answering prompts. These agents can monitor enterprise email flows, orchestrate cross-departmental supply chain logistics, resolve complex customer support tickets, and update SAP databases autonomously. The launch prompted widespread enterprise adoption discussions as Fortune 500 companies race to automate routine administrative workflows.

AI pioneer Dr. Fei-Fei Li announced the official launch of World Labs, a high-profile spatial intelligence startup that raised over $230 million to develop Large World Models (LWMs). Rather than generating flat 2D videos, World Labs' models understand the 3D physics, geometry, and affordances of real-world environments, allowing AI agents and robots to perceive, navigate, and manipulate spatial reality. The announcement ignited widespread discussions across robotics and computer vision on why true physical AI requires spatial understanding rather than pure language prediction.

Voice AI leader ElevenLabs launched its Conversational Voice AI Platform, allowing developers to deploy hyper-realistic, low-latency conversational voice agents across customer service, robotics, and interactive entertainment. With sub-second latency and realistic emotional cadence, the system handles real-time human interruptions, ambient background noise, and natural speech mannerisms smoothly. Demonstration videos of humanoid robots and interactive avatars holding fluid, empathetic voice conversations quickly went viral on YouTube and podcasts.

OpenAI previewed its upcoming 'o3' frontier reasoning model, demonstrating benchmark results that leapfrog current systems in competitive coding (Codeforces), high-school olympiad mathematics (AIME), and PhD-level science evaluations. Utilizing extended test-time compute and recursive self-verification loops, o3 solves intricate scientific proofs and algorithmic challenges that previously stumped frontier models. The announcement intensified industry discussions regarding test-time compute scaling laws and the rapid approach toward human-expert reasoning systems.

Engineers at the University of Cambridge unveiled a bio-inspired tactile electronic skin (e-skin) integrated with machine learning algorithms, enabling robotic grippers to perceive texture, slip, and pressure with human-grade sensitivity. In demonstration trials, robotic hands equipped with the e-skin successfully plucked ripe raspberries without bruising them, sorted slippery glass vials, and gripped fragile eggs with dynamic force modulation. The research is widely hailed as a critical missing link for scaling dexterous robots in agriculture, food packaging, and domestic chores.

Stanford University researchers released Mobile ALOHA 2, an upgraded, open-source bimanual teleoperation system designed to accelerate imitation learning for mobile robotic manipulation. Building on the viral success of the original Mobile ALOHA, version 2 features more robust mechanical linkages, upgraded 3D camera rigs, and improved ergonomics. Community videos showing the robot autonomously frying pans of food, washing dishes, and navigating hallways using lightweight consumer hardware continue to inspire thousands of robotics researchers and hobbyists worldwide.

Tokyo-based AI research lab Sakana AI, alongside international academic researchers, introduced 'The AI Scientist', the first comprehensive system for fully automated scientific discovery powered by foundation models. The system independently generates research hypotheses, writes code, executes experiments, visualizes results, authors complete scientific papers, and runs automated peer reviews. The concept of autonomous AI systems executing scientific discovery cycles sparked fierce debate across academia and tech media regarding the future of automated scientific breakthroughs.