Building the representation layer for enterprise AI
TieSet, Inc. was founded on the belief that AI systems can't reach their full potential without a persistent, evolving understanding of the entities they serve. STADLE is how we make that understanding real.
AI that doesn't understand entities will never outperform AI that does
Every enterprise AI system we've seen faces the same structural problem: the data about each customer, vehicle, asset, or user is fragmented across systems, frozen in time, and siloed by regulation. Models built on this data are perpetually incomplete.
STADLE exists to fix the data problem at its source — not by centralizing everything, but by learning a shared representation that carries the signal without carrying the raw data. The result is AI that actually knows who it's talking to.
Founded
TieSet, Inc. was founded in Silicon Valley. The founding team developed the core architecture behind STADLE, validated key algorithms in collaboration with KDDI Research, and established TieSet's foundational intellectual property.
First Enterprise Pilots
Enterprise pilots validated STADLE's representation infrastructure across industrial use cases. Over 300 users tested the platform, and the founders published Federated Learning with Python — reaching #5 on Amazon's ML category ranking.
Full Product Release
Full product release of STADLE as a continuously updating representation layer. Onboarded enterprise customers from architecture review through proof-of-concept deployment.
Built by researchers who shipped the technology
Kiyoshi Nakayama
Founder & CEO
Kiyoshi's research in federated learning, adaptive model aggregation, and distributed AI systems shaped the technical direction behind STADLE. He built his first distributed learning system at NEC Labs America and continued the work at Fujitsu Laboratories before founding TieSet to commercialize the technology.
George Jeno
Co-Founder & CTO
George leads STADLE's platform engineering — from the core aggregation protocol to the ModelOps Server and agent runtime. He brings deep experience in distributed systems and production ML infrastructure, ensuring STADLE performs at enterprise scale.
Organizations that need AI to keep understanding
KDDI Research
Telecommunications R&D
The research arm of KDDI Corporation, Japan's second-largest mobile carrier. KDDI Research operates at the frontier of network AI — exploring how distributed intelligence can extract signal from telemetry without centralizing subscriber data.
How they use STADLE
Federated learning infrastructure for distributed network intelligence, validating STADLE's core aggregation architecture across carrier-scale deployments.
Macnica
Technology Solutions & Distribution
One of Japan's leading technology distributors, bridging semiconductor and AI platforms with enterprise deployments across manufacturing, automotive, and logistics sectors.
How they use STADLE
Embedding STADLE as the representation layer within enterprise AI solutions delivered to Macnica's industrial customer base.
Nippon Life
Life Insurance
Japan's largest life insurer, managing long-horizon relationships with tens of millions of policyholders across distributed agent networks — where privacy constraints make centralized AI architectures unworkable.
How they use STADLE
Continuously updating member representations across policy lifecycle events without centralizing sensitive personal data across agent or branch boundaries.
Denso Wave
Industrial Automation & IoT
A DENSO Group company and inventor of the QR code, Denso Wave builds industrial IoT systems and factory automation solutions deployed across global manufacturing operations.
How they use STADLE
Edge-native AI for industrial equipment — device behavior representations that adapt to operational patterns without transmitting raw factory data upstream.
HIS
Travel & Tourism
One of Japan's largest travel agencies, operating across Asia with millions of customer touchpoints spanning tour packages, booking, and travel support services.
How they use STADLE
Customer intelligence that evolves across booking, planning, and post-trip touchpoints — personalization reflecting current intent, not past transactions.
Backed by institutions that back deep tech
Japan Innovation Campus
JIC · METI-sponsored · Palo Alto
TieSet is one of 5 startups selected as a resident member of Japan Innovation Campus (JIC) — a METI-sponsored startup hub in Palo Alto operated by Mori Building, connecting Japanese deep tech companies with the Silicon Valley ecosystem.
Plug and Play Japan
Accelerator Program · Winter 2026 Batch · Enterprise & AI
TieSet has been selected as a U.S. startup participant in Plug and Play Japan's Accelerator Program, Winter 2026 Batch / Enterprise & AI. We look forward to connecting with companies across Japan.
TieSet, Inc.
Headquarters
214 Homer Ave, Palo Alto, CA 94301
Japan Office
1-2-20 Kaigan, Minato-ku, Tokyo 105-0022, JAPAN
Founded
2020
General Inquiries
What we're building toward
“A world where every AI system starts knowing who it's serving — and keeps learning as that customer, vehicle, asset, or user changes.”