TraceWeave, a South Korean on-device AI startup developing the Phase-TITAN architecture, raised $2.1 million in seed funding.
South Korean on-device AI startup TraceWeave secured approximately $2.1 million in seed funding led by Mirae Asset Capital, with participation from QUAD Ventures, NAVER D2SF and Antler Korea. The company develops Phase-TITAN, an AI architecture designed for resource-constrained devices.

TraceWeave, a Seoul-based startup developing on-device artificial intelligence technology, raised approximately KRW 3 billion ($2.1 million) in a seed funding round led by Mirae Asset Capital. The round also included participation from QUAD Ventures, NAVER D2SF and Antler Korea.

The financing valued the company at approximately KRW 30 billion ($21.1 million) on a pre-money basis. TraceWeave plans to use the capital to expand its research and engineering teams, develop its AI architecture and support business growth.

Founded in October 2025, TraceWeave was established by a team including Jaeyoung Park, co-founder and CEO, TaeHoon Kim, co-founder and CTO, and Keunchul Park. The company emerged from Antler Korea’s seventh cohort and previously received pre-seed backing from the investor.

TraceWeave develops Phase-TITAN, a post-Transformer AI architecture designed to support complex AI tasks on devices with limited computing resources. Unlike conventional Transformer models, which typically require increasing key-value cache memory as context length grows, Phase-TITAN is designed to maintain a fixed-size context state while processing longer sequences of information.

The architecture is intended to enable AI agents to execute multi-step tasks, interact with tools, interpret outputs and determine subsequent actions directly on devices without relying on cloud-based inference.

TraceWeave has demonstrated a multi-step AI agent operating on a smartphone using its CPU without cloud calls. The company also completed a paid proof of concept with the autonomous driving subsidiary of a global automotive group, applying its technology to identify useful driving-data frames for improving autonomous driving models.

Additionally, TraceWeave is evaluating the performance of an in-vehicle AI agent with an automotive electronics software company. It is exploring further applications across robotics, drones, smart glasses and wearable devices.

The startup was also selected among the top 20 finalists of the NVIDIA Inception Grand Challenge 2026.

Get the latest startup funding news in your inbox.

We don’t spam! Read our privacy policy for more info.

Share