🤖 The Robot Beat

Tuesday, September 29, 2026

20 stories · Deep format

Generated with AI from public sources. Verify before relying on for decisions.

🎧 Listen to this briefing or subscribe as a podcast →

Semiconductor makers are pulling spatial intelligence directly into their silicon roadmaps. AMD's $8.2 billion acquisition of World Labs anchors today's briefing, signaling a push to fuse world-model software with edge hardware. Across the rest of the stack, developers are moving to untether foundation policies from cloud compute, releasing open-source engines and dedicated safety chips to run models securely on-device.

Humanoid Robots

MirrorMe Debuts VIVA Dexterous Hand and CADA Model for Real-Time Piano Duets

MirrorMe Technology unveiled its VIVA dexterous hand alongside the CADA music foundation model on Tuesday, September 29, performing a live real-time human-robot four-hand piano duet. The VIVA hand utilizes a tendon-linkage hybrid transmission that achieves joint speeds exceeding 1,000° per second and peak fingertip forces of 33 newtons. The accompanying CADA model processes acoustic and kinematic feedback in real time to coordinate expressive, contact-precise keystrokes.

Piano performance requires an extreme balance of high joint velocity, subtle force resolution, and closed-loop temporal feedback. Achieving human-level speeds and forces in a compact end effector validates tendon-linkage hybrid drives as a viable path for high-dexterity manipulation. This combination provides a strong benchmark for humanoids tackling intricate assembly tasks.

MirrorMe engineers assert that their hybrid architecture successfully resolves the traditional trade-offs between force, speed, and compact volume. Conversely, hardware reviewers point out that maintenance and tendon-wear over extended operating cycles remain critical factors for industrial adoption.

Verified across 1 sources: EQS News (Sep 29)

Newbility Unveils Billie Hybrid Humanoid with Wheeled Base and Articulated Torso

Autonomous delivery firm Newbility unveiled 'Billie' on Tuesday, September 29, a hybrid humanoid combining a four-wheeled mobile base with a dual-arm articulated torso. The robot features a 15 kg lifting capacity per arm and a 60 kg overall payload capacity. Newbility is scheduling proof-of-concept logistics trials with CJ CheilJedang in late 2026, leveraging autonomous navigation datasets gathered from its 400 existing delivery droids to train a unified spatial AI model.

Bipedal locomotion carries high energy costs and mechanical complexity that can hinder near-term commercial ROI in structured environments like warehouses and stores. By mounting articulated manipulation arms onto a proven wheeled platform, Billie delivers immediate material-handling utility without bipedal stability risks. This hybrid approach offers a pragmatic bridge for enterprise logistics automation.

Newbility leadership views the hybrid form factor as the fastest path to commercial profitability, backing it with data from hundreds of deployed sidewalk droids. Traditional biped developers argue, however, that wheeled bases restrict deployment on stairs and uneven terrain.

Verified across 1 sources: Maeil Business Newspaper (Sep 29)

AgiBot Deploys 100 Lingxi X2 Humanoid Robots Across Chinese Retail Chain

AgiBot deployed 100 Lingxi X2 humanoid robots across 100 retail locations owned by cookware manufacturer ASD in Wenzhou, Shanghai, and Wuxi on Monday, September 28. Participating in a three-day livestreamed promotion, the humanoids worked 10-hour continuous shifts handling greeting duties, answering product questions based on household needs, and executing sales handoffs.

Transitioning from single-unit pilot tests to a coordinated 100-unit commercial fleet deployment tests operational uptime and fleet management in public spaces. Running 10-hour daily shifts live on camera serves as a rigorous field endurance test for AgiBot's thermal management, joint reliability, and vision systems. Success here provides a template for expanding humanoids into customer-facing retail environments.

AgiBot executives frame the deployment as proof of commercial readiness and operational reliability. Retail analysts suggest that while greeting and sales handoffs generate foot traffic, proving long-term labor cost displacement remains necessary for wide adoption.

Verified across 1 sources: Yangtzeer (Sep 28)

Tesla Scales Optimus Assembly to Hundreds Weekly Amid Hand Assembly Bottlenecks

Following our report yesterday that Tesla scaled Optimus assembly at Fremont to several hundred units per week, new manufacturing details highlight persistent bottlenecks in building the Gen 3 hand. The complex appendage, featuring 22 degrees of freedom and over 100 components, requires tight tolerances that are limiting automated throughput—reportedly revising late-2026 production targets down to 1,000 weekly units from the 2,500-unit goal we tracked earlier. Additionally, worker reluctance to wear data-collection suits has forced Tesla to relocate motion-capture training to dedicated off-site facilities.

Tesla's hand assembly friction highlights the engineering gap between building automotive frames and mass-producing high-density, highly articulated robot limbs. Tight tolerances and thermal constraints in small actuators continue to require manual rework, limiting automated line throughput. These operational realities show that hardware manufacturing precision remains a primary gatekeeper for humanoid scaling.

Tesla manufacturing leads remain committed to aggressive weekly production targets backed by supply chain audits in Ningbo. Industrial manufacturing engineers counter that high component counts in 22-DoF hands will continue to bottleneck automated assembly lines without structural design simplifications.

Verified across 3 sources: Tech Times (Sep 28) · ScienceTimes (Sep 29) · RobotsBeat (Sep 28)

Consumer Robotics

Roborock Captures 23.7% Share to Lead Global Robot Vacuum Market in Q2 2026

Market data published by IDC on Tuesday, September 29, reveals that Roborock led global robot vacuum shipments in Q2 2026 with a 23.7% market share, up 2.9 percentage points year-over-year. Overall home cleaning robot shipments grew 21.5% to 11.2 million units, generating $3.89 billion in global revenue. Industry consolidation remains high, with five brands—Roborock, Dreame, ECOVACS, Xiaomi, and MOVA—controlling 75% of unit shipments and 82% of total revenue.

A 7.9% increase in average selling prices across the sector demonstrates that consumer demand is shifting toward premium, feature-rich models equipped with extendable robotic arms and obstacle-avoidance AI. Major manufacturers are successfully converting technical mechanical features into higher margin hardware sales. Market concentration among top players makes it increasingly challenging for low-cost generic vacuum brands to compete.

IDC analysts attribute Roborock's top market position to rapid deployment of high-end features like extendable side mops and high-suction bases. Consumer electronics strategists note that sustained growth will rely on expanding into adjacent lawncare and home-service form factors.

Verified across 3 sources: Home&Smart (Sep 28) · Aju Press (Sep 29) · MK (Sep 29)

Open-Source Robotics

Enactic Open-Sources $6,500 OpenArm Humanoid Bimanual Platform

Enactic, Inc. published OpenArm on GitHub on Monday, September 28, offering a fully open-source 7DOF humanoid arm designed for physical AI research. Featuring human-scale proportions and high backdrivability, a complete two-arm bimanual setup costs $6,500 USD and supports teleoperation, imitation learning, and simulation workflows. The repository also includes OpenArm Cell, a standardized physical enclosure aimed at improving benchmark reproducibility across academic and corporate labs.

High hardware costs have long restricted dexterous manipulation research to well-funded institution labs. By providing an open-hardware 7DOF arm design at $6,500, Enactic drastically lowers the entry barrier for physical AI experimentation. Standardizing the testing cell hardware also solves reproducibility challenges across different research teams.

Open-source robotics contributors on GitHub praised the project's accessible bill of materials and backdrivable design. Hardware engineers caution that long-term gear durability under continuous heavy loading will need further community testing.

Verified across 3 sources: Adafruit Blog (Sep 28) · GitHub (Sep 28) · Hacker News (Sep 28)

Robot AI

Black Forest Labs' 7B FLUX 3 Action Outperforms NVIDIA Models on RoboLab Benchmark

Black Forest Labs released FLUX 3 Action, a 7-billion-parameter open-weights World Action Model that jointly predicts future video frames and robot actions. Evaluated on NVIDIA's RoboLab-120 benchmark on Tuesday, September 22, the model secured the top position with a 42.9% success rate, outperforming NVIDIA's 16B Cosmos3-Nano-Policy and 3B GR00T N1.6. Notably, 19.55% of the model's action midtraining mixture utilized PC gameplay recordings, while third-party testing on a real Franka arm across ten DROID tasks yielded 28 successful runs out of 30 attempts.

The success of FLUX 3 Action illustrates how visual generative models trained on non-robotic datasets—such as video games and egocentric video—can effectively transfer to physical robotic manipulation. This provides a compelling alternative to the scarce, expensive datasets collected via manual teleoperation. For startup founders, high-performing open-weights models dramatically lower the compute costs associated with running proprietary closed-source policies.

Positronic Robotics demonstrated the model's real-world cost efficiency during physical Franka arm trials. Meanwhile, robotics researchers note that while open-weights models reduce reliance on cloud APIs, fine-tuning action-space parameters for custom hardware still presents integration hurdles.

Verified across 1 sources: Mixed (Sep 28)

Open-Source APXInf Engine Cuts Edge VLA Model Latency by 10.7x on Thor Silicon

Researchers from Tsinghua University, Wuwen Xinqiong, and Shanghai Jiao Tong University open-sourced APXInf, a full-stack edge inference engine tailored for embodied foundation models. Disclosed on Tuesday, September 29, the engine optimizes execution paths on NVIDIA Thor chips using FP8 precision, CUDA Graph, and Autotune. The setup reduced inference latency for the π0.5 model from 278ms to 26ms, achieving a 10.7x speedup and enabling a 38.46Hz real-time control frequency.

High inference latency on edge silicon has long limited complex vision-language-action policies from executing smooth, high-frequency motor control. By pushing execution speed past 38Hz directly on onboard hardware, APXInf removes a major computational bottleneck for untethered robots. This open-source tool allows small robotics teams to deploy large foundation policies without investing months in custom CUDA optimization.

The research team emphasizes that Rust-based memory safety and automated CUDA tuning make model integration repeatable. On the other hand, edge deployment engineers caution that maintaining peak FP8 performance under thermal throttling in mobile platforms remains a field challenge.

Verified across 1 sources: KuCoin (Sep 29)

Daimon Robotics Releases Tactile World Model Daimon-TWM for Contact Manipulation

Shenzhen startup Daimon Robotics launched Daimon-TWM on Tuesday, September 29, a tactile-grounded world model designed to process force, friction, and slip in real time. Building on its VTLA framework and proprietary vision-based tactile sensors, the model incorporates tactile interaction into motor decision loops for delicate assembly tasks. The rollout follows several funding rounds this year, including strategic backing from Ant Group.

While vision-language-action policies excel at spatial path planning, they often struggle during contact-rich tasks where subtle force shifts dictate success. Daimon-TWM elevates touch from passive safety monitoring to an active input for policy generation. Incorporating rich tactile data helps resolve the 'physical drift' that frequently causes purely vision-based policies to fail in unstructured environments.

Daimon Robotics highlights its integrated sensor-to-model data pipeline as essential for mastering complex assembly work. Industry observers note that scaling tactile policies will require industry-wide standardization of tactile sensor hardware and data formats.

Verified across 1 sources: Robotics and Automation News (Sep 29)

SAIL Framework Uses Search-Time Compute to Lift VLA Trajectory Success to 73%

Researchers from the University of Tokyo and Sakana AI introduced SAIL on Monday, September 28, a method that applies Monte Carlo Tree Search to vision-language model planning. Tested on Gemini Robotics-ER 1.5 across six ALOHA simulation tasks without fine-tuning weights, spending extra search compute at inference increased trajectory generation success from 25% (single pass) to 73% (45 search nodes). The team validated the approach on a physical LeRobot SO-101 arm, achieving five successful block placements out of six trials.

Applying inference-time compute scaling—a technique widely used in reasoning LLMs—to physical robot trajectories offers a way to improve policy execution without massive dataset fine-tuning. This approach allows developers to dynamically scale computation budgets based on task complexity. Furthermore, successful search trajectories can be saved to create a high-quality data flywheel for downstream imitation learning.

Sakana AI researchers demonstrate that simulation search successfully filters out execution failures before physical deployment. Robotics practitioners note that while search compute improves success rates, the extra latency per action currently restricts its use to non-time-critical manipulation.

Verified across 3 sources: Complete AI Training (Sep 28) · Runtime Wire (Sep 28) · X (Sep 27)

Robotics Tech

Sharpa Debuts D01 Tactile Robot and AE01 Haptic Exoskeleton Glove at IROS 2026

Robotics hardware developer Sharpa launched three tactile platforms at IROS 2026 in Pittsburgh on Tuesday, September 29: the D01 integrated tactile robot, the W02 compact hand, and the AE01 haptic exoskeleton glove. The D01 pairs 7-DOF arms and full-body electronic skin with a 1:1 payload-to-weight ratio operating over a 1,000 Hz communication bus. The AE01 glove provides 22 degrees of freedom and 256 levels of vibrotactile feedback to enable precise teleoperation and training data capture.

High-frequency communication combined with full-body electronic skin addresses critical hardware limitations in delicate contact tasks. Providing 256 levels of haptic feedback via an exoskeleton glove equips operators to record high-fidelity tactile datasets during teleoperation. This hardware suite helps bridge the gap between human sensory perception and machine motor learning.

Sharpa engineers highlight that 1,000 Hz communication speeds eliminate control latency during fast contact tasks. Academic researchers note that while full-body e-skin improves safety and feedback, long-term sensor durability against workplace abrasion requires continued validation.

Verified across 1 sources: Siam News (Sep 29)

Saginaw Precision Launches Michigan Facility for Domestic Robotic Ball Screws

Salient Motion announced the launch of Saginaw Precision on Monday, September 28, establishing an in-house manufacturing plant in Saginaw, Michigan, for precision ball and roller screws. Led by Garrett O'Brien, the SP-1 facility utilizes modern machining and software to reduce custom prototype actuator lead times from years down to eight weeks, projecting an annual production capacity of 50,000 ball screws by 2027.

Precision ball and roller screws are critical components for high-force actuators in humanoid joints and industrial automation, yet domestic production capacity has historically suffered from multi-year supply chain bottlenecks. Vertically integrating production in Michigan directly targets this component lead-time barrier for North American hardware developers. Faster domestic component prototyping speeds up hardware iteration cycles for the entire robotics industry.

Salient Motion executives emphasize that localized, software-managed machining is key to drastically cutting prototype lead times. Supply chain analysts agree that domestic actuator component supply reduces geopolitical risk, though scaling manufacturing quality to match established overseas suppliers will take time.

Verified across 1 sources: ManufacturingTomorrow (Sep 28)

Robotics Startups

Faraday Future Spins Off Robotics Division into $200 Million Public Entity

AIxCrypto Holdings (NASDAQ: AIXC) entered a non-binding agreement on Monday, September 28, to acquire Faraday Future's robotics business for approximately $200 million in stock. Under the agreement, AIxC will exit its cryptocurrency strategy, rename itself FF EAI Robotics Ecosystem Inc., and begin trading under ticker FFR on September 30. Faraday Future's robotics division reported 552 unit shipments through August 2026 across its ecosystem.

This transaction highlights a corporate pivot away from cryptocurrency management into a public pure-play physical AI entity. Spinning out the robotics division provides Faraday Future with an independent public equity currency to fund hardware development without diluting its core automotive operations. However, achieving projected multi-year revenue growth will depend heavily on scaling hardware shipments beyond initial low-volume runs.

Faraday Future leadership views the spin-off as a vehicle to unlock standalone public valuation for its embodied AI stack. Financial analysts urge caution, pointing out that reverse-merger structures face intense public market scrutiny if shipment volumes fail to scale rapidly.

Verified across 2 sources: TradingView (Sep 28) · The Automotive Data (Sep 29)

IPercept Closes $16.5 Million Series A for Controller-Agnostic CNC Machine Diagnostics

Stockholm-based startup IPercept closed a $16.5 million Series A round on Tuesday, September 29, jointly led by Isogon Ventures and 2150. Founded in 2019, IPercept provides micrometer-level motion diagnostic systems for industrial CNC machine tools that operate without connecting to internal machine controllers. The company currently serves over 30 enterprise clients—including Airbus, Bosch, and Scania—and plans to use the capital to expand into the U.S. market.

While humanoids capture significant venture attention, traditional industrial CNC machine parks represent a massive asset base operating at low average efficiency due to unplanned downtime. IPercept's controller-agnostic sensor approach bypasses complex integration hurdles by measuring physical mechanical motion directly. Securing backing from industrial giants validates the demand for non-invasive predictive maintenance across heavy manufacturing.

IPercept leadership argues that domain-specific mechanical modeling offers faster implementation than generic black-box AI diagnostics. Manufacturing operations directors support non-invasive sensing because it avoids voiding original equipment manufacturer controller warranties.

Verified across 1 sources: Tech Funding News (Sep 29)

Healthcare Robotics

FDA Releases Updated Draft Guidance Standardizing Surgical Robot Premarket Submissions

Following the 50-page FDA draft guidance for surgical robots we covered last week, the agency has announced a two-day public workshop scheduled for December 2–3 to evaluate autonomous and telesurgical robotics. While the new framework clarifies premarket requirements for practitioner-controlled systems, the upcoming workshop aims to tackle the distinct regulatory challenges posed by independent and remote-controlled surgical platforms.

While the initial draft guidance clears up premarket ambiguity for practitioner-controlled platforms, the December workshop signals the FDA's intent to establish explicit regulatory pathways for autonomy and remote teleoperation—key hurdles for the next generation of healthcare robotics.

Medtech analysts at Evercore ISI praised the draft guidance for providing a clear, standardized approval roadmap that removes historical software submission ambiguity. Regulatory consultants note that excluding autonomous and telesurgical systems keeps immediate focus on practitioner-driven platforms while setting up future regulatory frameworks in December.

Verified across 4 sources: Cardiovascular Business (Sep 28) · MedTech Dive (Sep 28) · Dig.Watch (Sep 28) · Health Recovery Support (Sep 28)

AI Hardware

AMD Acquires Fei-Fei Li's World Labs for $8.2 Billion to Expand Spatial AI Silicon

AMD has agreed to acquire World Labs, the spatial-intelligence startup co-founded by Fei-Fei Li, in an all-stock transaction valued at $8.2 billion. Announced on Monday, September 28, the deal brings Li to AMD as Executive Vice President and Chief Scientist to shape future chip architectures around spatial world models. World Labs previously introduced its Atlas spatial model earlier in September, and the transaction is expected to close by the end of 2026 pending regulatory approval.

This transaction represents a major strategic move by AMD to build a full-stack physical AI ecosystem capable of competing directly with NVIDIA's Omniverse and Isaac platforms. By incorporating spatial intelligence model research into its silicon development, AMD is positioning its future GPU and accelerator architectures to natively handle 3D scene generation and physical simulation. For robotics entrepreneurs, closer integration between world-model software and silicon could yield dedicated edge hardware optimized for spatial inference.

AMD leadership presents the acquisition as a necessary step to lead in spatial computing and physical AI workloads. However, market investors expressed immediate caution over the early commercial stage of World Labs, driving AMD stock down nearly 4% following the announcement.

Verified across 2 sources: Tech Insider (Sep 29) · TradingKey (Sep 29)

NVIDIA Debuts Open Agent Safety Platform with BlueField-4 Hardware Enforcement

NVIDIA launched the Open Agent Safety Platform on Tuesday, September 29, pairing an open-source secure runtime called OpenShell with a hardware reference architecture named Sentry. Executing on NVIDIA BlueField-4 DPUs, Sentry provides in-silicon zero-trust enforcement to monitor agent requests and isolate anomalous actions within milliseconds. Humanoid and physical AI companies including Figure, Gecko Robotics, and Skild AI are collaborating with NVIDIA to integrate these hardware-isolated safety boundaries into their autonomous platforms.

As embodied foundation models gain broader execution capabilities, relying strictly on prompt guardrails is insufficient to prevent dangerous physical actions. Moving safety controls into dedicated hardware domains like DPUs creates an unalterable firewall between neural network outputs and motor actuation. This architecture sets a crucial standard for deploying autonomous humanoids in shared human environments.

NVIDIA CEO Jensen Huang and Figure CEO Brett Adcock highlighted that independent hardware runtimes are mandatory as humanoids scale into homes and factories. However, integration details remain open, as engineering teams must balance strict safety interrupts against real-time motor control responsiveness.

Verified across 2 sources: Humanoids Daily (Sep 28) · Rocking Robots (Sep 29)

SiMa.ai Secures $150 Million Series C to Build 1,000 TOPS Modalix Physical AI Chip

Silicon startup SiMa.ai closed a $150 million Series C round co-led by Fidelity and Amplify on Monday, September 28, bringing its valuation to $1.45 billion. The capital will support the expansion of its Palette Neat software suite and the development of Modalix, its next-generation system-on-chip targeting 1,000 TOPS of compute for physical AI edge systems. Modalix is scheduled for commercial release in the first half of 2028.

Scaling physical AI requires edge processors that deliver high TOPS per watt without relying on continuous cloud connectivity or heavy liquid cooling. SiMa.ai's focus on low-power, high-throughput edge execution addresses a major hardware constraint for autonomous drones, mobile robots, and humanoids. This capital injection underscores strong investor appetite for alternative silicon architectures that challenge NVIDIA's edge dominance.

SiMa.ai executives position Modalix as a cost-effective, purpose-built chip tailored specifically for multimodal physical inference. Market analysts note, however, that competing against NVIDIA's established CUDA software ecosystem remains a steep hurdle for early-stage chipmakers.

Verified across 1 sources: aiunderstanding.org (Sep 28)

Microrobotics

MIT Engineers Design Muscle-Powered Swimming Microrobot Controlled by Light

Engineers at MIT, led by Ritu Raman, developed a paper-thin, muscle-powered swimming robot, published on Tuesday, September 29. Measuring 0.5 mm thick, the robot consists of a GelMA hydrogel skeleton covered with genetically engineered muscle cells that twitch in response to light. By flashing light onto its twin fins at distinct intervals, researchers steered the microrobot through an aquatic maze at speeds up to four body lengths per minute while optimizing muscle alignment using stamped micro-grooves.

Biohybrid microrobotics frequently suffer from heavy 3D tissue requirements that restrict movement and fluid flow. MIT's 2D planar design achieves high force-to-weight efficiency while retaining the self-healing advantages of living muscle tissue. This approach offers a foundation for developing soft, bio-compatible micro-swimmers for targeted biomedical or environmental sampling.

The MIT research team emphasizes that micro-grooved gel skeletons maximize biological force output without rigid structural weight. External biophysicists point out that sustaining biohybrid muscle viability outside controlled nutrient media remains a key challenge for long-term field use.

Verified across 1 sources: MIT News (Sep 29)

Soft Robotics

Machine Learning Optimizes 3D-Printed Soft Pneumatic Gripper to 600% Stretchability

Building on the ML-formulated, 600%-stretchable 3D-printing resin from South Korean researchers that we've been tracking, the team has now demonstrated an 11-segment soft pneumatic gripper. Decoupling high-viscosity resin properties from printer flow constraints allowed them to successfully print the actuator using only 0.5 mL of material, which was then used to pick up delicate objects like raw eggs without damage.

Successfully applying the AI-formulated resin to a multi-segment gripper validates the material's viability for delicate physical handling. This methodology accelerates the rapid prototyping of custom soft robotic end effectors without the structural defects commonly caused by high-viscosity elastomers.

The Korean development team notes that predictive simulation eliminates trial-and-error iterations in resin formulation. Soft robotics manufacturing specialists add that scaling this ML formulation process to multi-material industrial printers will be necessary for commercial adoption.

Verified across 1 sources: Robot Today (Sep 28)


The Big Picture

Semiconductor Giants Consolidate Spatial AI Software AMD's $8.2 billion acquisition of World Labs demonstrates that hardware vendors are looking beyond raw compute to capture the spatial generation and 3D simulation software layers essential for physical AI.

Edge Inference Optimization Opens Real-Time VLA Control Engine frameworks like APXInf achieve sub-30ms latencies for models such as π0.5 on edge chips, enabling open-weights action policies to run on local hardware without cloud round-trips.

Tactile Sensing Integrates Into World Action Models Developments from Daimon Robotics and Sharpa reflect an industry shift toward embedding real-time force, slip, and haptic feedback directly into decision-making foundation models.

Open Hardware Arms Lower Physical AI Barriers Releases like Enactic's OpenArm and Seeed's Isaac ROS 5.0 integration provide affordable, standardized bimanual platforms that standardize research and speed up industrial testing.

Hybrid Locomotion Bridges Commercial Industrial Gaps Newbility's Billie and existing fleet rollouts highlight how pairing wheeled bases with articulated upper bodies offers immediate commercial productivity over unproven bipedal endurance.

What to Expect

2026-11-24 — FDA public comment window closes for autonomous and telesurgical robotics draft guidance.
2026-11-30 — Open Robotics 2026 Community Survey submission deadline.
2026-12-02 — FDA two-day public workshop on autonomous and remote-operated surgical robotics.
2026-12-31 — Expected closing window for AMD's $8.2B acquisition of World Labs.

Every story, researched.

Every story verified across multiple sources before publication.

🔍

Scanned

Across multiple search engines and news databases

509
📖

Read in full

Every article opened, read, and evaluated

149
⭐

Published today

Ranked by importance and verified across sources

20

— The Robot Beat

🎙 Listen as a podcast

Subscribe in your favorite podcast app to get each new briefing delivered automatically as audio.

Apple Podcasts
Library tab → ••• menu → Follow a Show by URL → paste
Overcast
+ button → Add URL → paste
Pocket Casts
Search bar → paste URL
Castro, AntennaPod, Podcast Addict, Castbox, Podverse, Fountain
Look for Add by URL or paste into search

Spotify isn’t supported yet — it only lists shows from its own directory. Let us know if you need it there.