Summary
Sunday Robotics is a Mountain View, California-based startup building Memo, a wheeled home robot designed to autonomously perform household chores such as clearing tables, loading dishwashers, folding laundry, and even pulling espresso shots. Founded in late 2024 by Stanford robotics researchers Tony Zhao and Cheng Chi, Sunday differentiates itself from competitors by avoiding in-home teleoperation: instead of controlling robots remotely inside customers’ homes, the company trains Memo using data collected by a distributed network of human “Memory Developers” who perform everyday tasks while wearing a proprietary Skill Capture Glove that mirrors the robot’s own hand and sensor layout.
The company emerged from stealth in November 2025 with a $35 million Series A led by Benchmark and Conviction, then reached unicorn status just four months later in March 2026 with an oversubscribed $165 million Series B led by Coatue Management at a $1.15 billion valuation—one of the fastest valuation climbs in the humanoid and home-robotics sector. Sunday has stated it trained Memo on approximately 10 million episodes of real household routines and plans to begin an in-home beta program with select “Founding Family” households in late 2026, targeting deployment by Thanksgiving 2026.
Zhao and Chi are known in the robotics research community as co-creators of some of the field’s most widely adopted open-source imitation-learning tools: ALOHA and ACT (Action Chunking with Transformers), developed by Zhao at Stanford, and Diffusion Policy and the Universal Manipulation Interface (UMI), developed by Chi. Sunday’s core technical bet is that combining large-scale, real-world human demonstration data with these imitation-learning techniques—rather than relying on simulation, teleoperation, or industrial robot retrofits—is the fastest path to a genuinely useful home robot.
Company Story
Tony Zhao and Cheng Chi first connected on social media after noticing each other’s research papers, published roughly a month apart. Zhao, then a Stanford PhD student advised by Chelsea Finn, had co-created ALOHA (A Low-cost Open-source Hardware system for bimanual teleoperation) and ACT (Action Chunking with Transformers), demonstrating that a $20,000 dual-arm robot system could learn precise manipulation tasks—threading zip ties, juggling ping-pong balls, assembling chains—from just minutes of human demonstration data. He later extended the work to Mobile ALOHA, a mobile bimanual platform that performed household-style tasks like sautéing shrimp and cleaning wine spills. Chi, who moved from a PhD program at Columbia to Stanford under advisor Shuran Song, had developed Diffusion Policy, which used denoising diffusion techniques to generate more robust robot action trajectories, and the Universal Manipulation Interface (UMI), a low-cost gripper system for capturing high-fidelity, transferable manipulation data.
In late 2024, the two researchers began working together informally, clamping a robot arm to a desk in Chi’s apartment and experimenting with simple household tasks. Zhao dropped out of his Stanford PhD program to co-found Sunday Robotics with Chi, and by the end of 2024 the initial two-person team had grown to roughly eight people. In December 2024, their prototype robot, later named Memo, had only one arm and had learned its first skill: arranging shoes.
Rather than pursuing a bipedal humanoid form factor or relying on in-home teleoperation—the approach used by some competitors—Sunday made two deliberate architectural choices. First, it built Memo as a wheeled, non-humanoid robot, on the philosophy that simplifying the hardware problem where possible would let the team focus engineering effort on the harder problem of generalizable manipulation. Second, it built a scalable data-collection pipeline centered on a low-cost (reportedly $200–$400) wearable glove system, called the Skill Capture Glove, that mirrors Memo’s own hand and sensor configuration. Rather than have engineers teleoperate robots directly inside customer homes, Sunday recruited a distributed network of U.S.-based “Memory Developers” who wear the gloves to perform ordinary household tasks—loading dishwashers, folding laundry, tidying—in their own homes, generating training data that maps human movement directly onto the robot’s action space.
Through 2025, Memo’s capabilities expanded steadily: by October 2025 it could fold piles of socks, handle delicate wine glasses, and pull a shot of espresso with visible crema. The company emerged from stealth in November 2025, announcing a $35 million Series A led by Benchmark and Conviction and unveiling Memo publicly for the first time, stating the robot had been trained on approximately 10 million episodes of household routines gathered from more than 500 real U.S. homes.
The public unveiling generated significant industry interest, and by March 2026—just four months later—Sunday closed an oversubscribed $165 million Series B led by Coatue Management, with Bain Capital Ventures, Fidelity Management & Research Company, Tiger Global, Benchmark, Conviction, and Xtal Ventures also participating. The round valued the company at $1.15 billion, making Sunday one of the fastest robotics startups to reach unicorn status, and Coatue co-founder Thomas Laffont joined the company’s board. Sunday said the new capital would fund a shift from public demonstrations to real-world deployment, with engineering headcount growing 3x, research 4x, and data operations 5x, in preparation for an in-home beta program slated for the fall of 2026.
In July 2026, Sunday unveiled ACT-2, an updated AI model powering Memo, claiming the robot could fold laundry successfully more than 99% of the time in homes it had never seen before, including on garment types it had not specifically been trained on—a claimed advance in cross-home generalization that the company positioned as a key unsolved problem in household robotics.
Founders
Tony Zhao (Co-founder and CEO) earned a B.S. in electrical engineering and computer sciences from UC Berkeley in 2021, studying under advisors Sergey Levine and Dan Klein, before joining Stanford’s PhD program in 2021 under Chelsea Finn at the IRIS Lab, where he received a Stanford Robotics Fellowship. During his PhD, Zhao published ACT (Action Chunking with Transformers) and ALOHA in 2023, followed by Mobile ALOHA in 2024—open-source contributions that became widely used benchmarks and hardware platforms across the robot-learning research community, including integration into Hugging Face’s LeRobot framework. He held internship and research roles at DeepMind and Tesla before dropping out of his Stanford PhD program in 2024 to co-found Sunday Robotics with Cheng Chi.
Cheng Chi (Co-founder and CTO) began his PhD at Columbia University before moving to Stanford to work under advisor Shuran Song. Chi is the lead author of Diffusion Policy, a widely cited framework applying denoising diffusion models to robot action generation, and developer of the Universal Manipulation Interface (UMI), a low-cost gripper system for scalable, high-fidelity robot demonstration data collection. His apartment served as the earliest workspace for Sunday Robotics, where he and Zhao first prototyped what became Memo.
Products
Memo is Sunday’s flagship home robot, a wheeled (non-bipedal) platform designed to autonomously perform a range of household chores without requiring live, in-home teleoperation. As of mid-2026, Memo’s demonstrated skill set includes clearing dinner tables of plates, utensils, and delicate glassware; disposing of food scraps; loading and running dishwashers; folding laundry, including socks; and operating an espresso machine to pull a shot with crema. The company says Memo is powered by an expanding “Skill Library” of AI models and continues to learn new skills on a roughly monthly cadence. Unlike some competitors that rely on remote human operators controlling the robot from inside a customer’s home during live use, Sunday trains Memo entirely from pre-collected human demonstration data, aiming for the robot to work “autonomously out-of-the-box” once deployed.
Skill Capture Glove is Sunday’s proprietary data-collection hardware, a wearable glove system that mirrors the shape, joint configuration, and sensor layout of Memo’s own hands. Reported to cost roughly $200 to $400 to produce, the glove is worn by a distributed network of more than 500 U.S.-based “Memory Developers” who perform everyday household tasks in their own homes while the glove records forces, grip patterns, and motion data. This human demonstration data is then used to train Memo’s manipulation policies, an approach Sunday argues produces more transferable, real-world-representative training data than simulation or teleoperation-based methods, while explicitly avoiding the privacy concerns associated with cameras operating live inside a paying customer’s home.
ACT-2 is Sunday’s most recent AI model, announced in July 2026, which the company says materially improved Memo’s ability to generalize learned skills to new homes and previously unseen objects. Sunday reported that Memo folded laundry successfully more than 99% of the time when tested in unfamiliar homes on garments it had not specifically been trained to handle, an outcome the company positioned as evidence that skills learned in one home can transfer to another without extensive site-specific retraining—one of the central technical challenges in deploying general-purpose home robots at scale.
Funding
Sunday has raised a total of $200 million across two disclosed institutional rounds in under five months, a notably rapid fundraising cadence even by the standards of the current robotics investment boom.
The company’s first disclosed institutional round was a $35 million Series A on November 20, 2025, led by Benchmark and Conviction, announced concurrently with Memo’s public unveiling after roughly a year of stealth development. According to industry analysis from Sacra, Sunday had also raised earlier, smaller pre-institutional capital as it transitioned from academic research toward a product company, though only the Series A and Series B rounds have been publicly disclosed with amounts and lead investors.
Four months later, on March 12, 2026, Sunday closed an oversubscribed $165 million Series B led by Coatue Management at a $1.15 billion post-money valuation, with Thomas Laffont, co-founder of Coatue, joining the company’s board. The round also included Bain Capital Ventures, Fidelity Management & Research Company, Tiger Global, Benchmark, Conviction, and Xtal Ventures. Coatue’s Laffont specifically cited Zhao’s research pedigree and the team’s “velocity” in translating research into a shipping product as key factors in the investment decision. Bain Capital Ventures partner Aaref Hilaly framed the investment thesis around Sunday’s full-stack, non-teleoperated approach, stating that “Sunday’s integrated approach moves past one-time, teleoperated demos and into a world where robots reliably serve humans every day.”
Partnerships
Sunday’s most distinctive operational structure is not a traditional corporate partnership but its Memory Developer network—a distributed group of more than 500 U.S.-based individuals who are compensated to wear the company’s Skill Capture Glove and perform everyday household tasks in their own homes, generating the demonstration data that trains Memo. The company has said it is on track to scale this data-operations function fivefold following its Series B, and frames the network as a core, defensible asset rather than a one-time data-collection exercise.
Sunday’s investor base doubles as a strategic network: Coatue Management, Bain Capital Ventures, Fidelity Management & Research Company, Tiger Global, Benchmark, and Conviction bring deep experience across consumer technology, AI infrastructure, and prior robotics investments, and board member Thomas Laffont’s involvement signals continued growth-stage support heading into Sunday’s planned 2026 commercial beta.
Ahead of general availability, Sunday has run public demonstrations of Memo operating in Airbnb rental properties the robot had not previously encountered, using these unfamiliar environments as informal proof points for its claims about cross-home generalization—an approach the company has discussed publicly as evidence for “zero-shot generalization” rather than a formal commercial partnership with Airbnb itself.
Timeline
2024 — Tony Zhao and Cheng Chi begin informally prototyping a robot in Chi’s apartment in late 2024. Zhao drops out of his Stanford PhD program to co-found Sunday Robotics. By December, the team’s single-arm prototype—later named Memo—learns its first skill, arranging shoes. The team grows to roughly eight people by year’s end.
2025 — Memo’s capabilities expand through the year, learning to fold socks, handle glassware, and pull espresso shots by October. In November, Sunday emerges from stealth, publicly unveiling Memo alongside a $35 million Series A led by Benchmark and Conviction, and disclosing that Memo had been trained on roughly 10 million episodes of household routines from more than 500 real homes.
2026 — In March, Sunday closes an oversubscribed $165 million Series B led by Coatue Management at a $1.15 billion valuation, reaching unicorn status roughly 15 months after founding. The company announces plans to shift from demos to real-world deployment, with a beta program targeted for households by Thanksgiving 2026. In July, Sunday unveils its ACT-2 model, reporting a greater than 99% success rate folding laundry in previously unseen homes.
Market Context
Sunday Robotics enters the home-robotics market at a moment of intense investor and public interest in bringing general-purpose robots into private homes, alongside competitors such as 1X Technologies (NEO), which similarly targets consumer households but has pursued a bipedal humanoid form factor and, controversially, relies in part on live in-home teleoperation for tasks beyond the robot’s current autonomous capability. Sunday’s decision to build a simpler, wheeled robot and to explicitly avoid live in-home camera-based teleoperation represents a direct strategic and philosophical counterpoint to that approach, and the company and its investors have publicly framed this as a key differentiator addressing both the reliability gap and the privacy concerns associated with teleoperated home robots.
The rapid pace of Sunday’s fundraising—from stealth emergence to unicorn status in roughly four months—reflects broader capital markets enthusiasm for embodied AI and home robotics in 2025-2026, a period that has seen large rounds across the sector, including 1X Technologies’ reported pursuit of a $1 billion raise at a $10 billion-plus valuation and Figure AI’s $1 billion Series C at a $39 billion valuation. Sunday’s $1.15 billion valuation, achieved with only $200 million raised and no commercial product shipped to paying consumers as of mid-2026, reflects investor confidence in the founders’ research pedigree and technical approach more than proven commercial traction.
Founders Tony Zhao and Cheng Chi bring unusually strong technical credibility to the venture: their prior open-source research (ALOHA, ACT, Diffusion Policy, and UMI) is widely used across the academic and industrial robot-learning community and has shaped how many other companies and labs approach the problem of learning manipulation skills from demonstration data. This research pedigree has been repeatedly cited by investors, including Coatue’s Thomas Laffont, as central to their investment thesis.
The company’s central technical and commercial bet—that off-site, glove-based human demonstration data can produce a home robot that works reliably “out-of-the-box” without in-home teleoperation—remains unproven at meaningful scale. Sunday’s own public claims, including the 99% laundry-folding success rate on unfamiliar garments and homes, are self-reported and have not yet been subjected to independent, large-scale verification. The company’s planned late-2026 beta program, deploying robots to a limited number of “Founding Family” households, represents the first real test of whether Memo’s demonstrated skills generalize reliably to the full diversity of everyday homes, or whether—as with earlier waves of home robotics attempts—the gap between an impressive demo and a dependable product proves difficult to close.
Pricing and general availability remain undetermined. Zhao has stated that internal prototypes currently cost between $6,000 and $20,000 to build using CNC machining and hand-finishing, but that a transition to injection-molded manufacturing could bring material costs below $10,000, which he has suggested would inform Memo’s eventual retail price. No confirmed consumer price or general commercial launch date had been announced as of mid-2026, with the company’s public messaging focused on the 2026 beta program rather than a broad retail launch.
Source Notes
This profile draws on Sunday Robotics’ own website and product announcements, GlobeNewswire press releases covering the Series A and Series B rounds, and independent reporting and analysis from TechCrunch, Business Insider, SiliconANGLE, Tech Funding News, Sacra, Tracxn, and Bain Capital Ventures (an investor in the company).
Funding figures, dates, and investor lists are drawn from Sunday’s own Series B announcement (“Sunday’s Series B: No More Demos”), corroborating GlobeNewswire press releases, and Tracxn’s funding database, all of which are consistent on the $35 million Series A (November 2025) and $165 million Series B (March 2026, $1.15 billion valuation) figures.
Founder biographical and research details—including Tony Zhao’s work on ALOHA, ACT, and Mobile ALOHA, and Cheng Chi’s work on Diffusion Policy and UMI—are corroborated across multiple independent sources including Bain Capital Ventures’ investor analysis, GreyJournal, sudoremove.com, and Sunday’s own public statements in interviews such as the No Priors podcast. Product capability claims, including the reported 99% laundry-folding success rate and the 10-million-episode training dataset, are self-reported by the company and have not been independently verified; they are presented in this profile as company claims rather than confirmed third-party findings. Given the fast-moving nature of the company’s fundraising, product development, and beta program timeline, figures should be re-verified against current sources.