Sixth Sense joins Y Combinator’s S26 batch and raises about $700,000

Sixense selected for Y Combinator’s U.S. Summer Batch… raises ₩1 billion in pre-seed funding

Physical AI & Robotic Multimodal Data

Tactile robotics AI platform Sixense selected for Y Combinator’s U.S. S26 batch… raises ₩1 billion in pre-seed funding

Former Tesla Vision Robotics team member joins… launches $1 million deal pipeline with synchronized visual and tactile data platform

Sixense, a startup building a tactile robotics AI training-data platform to lead the global physical AI data ecosystem, announced on the 8th that it was selected for Y Combinator’s 2026 Summer Batch (S26) and raised approximately ₩1 billion in pre-seed funding. Following an early investment anchor from Crew Capital, the company has earned recognition on the global stage for its exceptional hardware-software integration.

Sixense is an innovative deep-tech company co-founded by four experts in global big tech and robotics, led by CEO Jongjin Baek, a Georgia Tech alumnus. The team includes Matthew Wulff, who previously led a 20-person Vision Robotics team at Tesla. They identified a key pain point: existing robot training-data pipelines rely solely on visual information from 2D and 3D cameras, limiting their ability to recognize complex textures and achieve precise control.

ROBOTIC DATA PIPELINE: Sixense’s Multimodal Architecture and Track Record

Real-time visual-tactile synchronization

Combines in-house tactile gloves with first-person camera sensors to map hand mechanics and subtle pressure on the same time axis (time sync)

End-to-end data automation

Fully automates the entire process within its platform, from sensor calibration and data cleaning to quality validation and token labeling

Exceptional global market fit

Capturing data at global factories in Vietnam and India; secured a contract request worth $1 million after providing just one hour of sample data

Sixense’s core proprietary moat lies in its technology for precisely binding tactile context—including human gestures and resistance forces generated during manual work—to visual streams. The company is currently collecting data for foundation model training in field environments across global manufacturing hubs, including Vietnam, India, and China. After providing its first design partner with just one hour of input data, it received indications of interest in a proof of concept (PoC) and contracts worth at least $1 million (approximately ₩1.4 billion), demonstrating strong business value.

“In the era of Physical AI, which interacts with the real physical world rather than handling parameters in a virtual world like LLMs, a high-quality tactile data pipeline capable of enabling human-level manipulation—not just a video layer—will become the operating system (OS) and core infrastructure of the humanoid robotics ecosystem,” said Sixense CEO Jongjin Baek. “We plan to leverage Y Combinator’s global acceleration infrastructure to mass-produce our next-generation tactile glove hardware stack and aggressively scale runtime partnerships with global robotics and big tech companies.”

K-Valley View: Expert Analysis

Sixense’s selection for Y Combinator’s batch and launch of pre-seed funding underscore the robotics AI paradigm’s shift from vision-based locomotion to fine-grained multimodal manipulation. Just as Deepley targeted global supply chains with an architecture for sensing subtle sounds in manufacturing processes, Sixense has built a moat by binding “tactile context”—a critical missing link in real-world manufacturing data—through a time-synced flywheel design. The biggest challenge for humanoid robotics players such as Tesla Optimus is turning the subtle pressure and torque-control mechanisms of the human hand into data. Sixense has built a “data engine” that automates the entire pipeline, from sensor synchronization to token labeling, dramatically reducing friction in data collection. Triggering a $1 million deal pipeline in the North American manufacturing belt with just one hour of sample data suggests a strong potential for its data layer to become an exclusive runtime standard asset for global Physical AI factories.