NAVER D2SF Invests in Clone Labs, an AI Startup That Predicts User Intent and Decisions
NAVER D2SF Invests in Clone Labs, an AI Startup That Predicts User Intent and Decisions
NAVER D2SF Invests in Clone Labs, an AI Startup That Predicts User Intent and Decisions
-Developing a "User Model" that records and accumulates users' computer and agent usage patterns in real time to predict next actions
- NAVER D2SF continues to back founders identifying new opportunities emerging from the AI paradigm shift
May 18, 2026
NAVER D2SF, the corporate venture capital arm of NAVER, has made a new investment in Clone Labs(CEO, Minchan Kim), an AI startup focused on understanding user intent and context. Clone Labs develops a User Model that learns computer and agent usage patterns to predict and execute next actions on behalf of users. NAVER D2SF decided to invest in the team based on its ability to quickly identify emerging problems in the AI era and pursue them relentlessly through experimentation.
As AI agents become more widely adopted, the very process of issuing commands and reviewing results is emerging as a new bottleneck. Paradoxically, the faster agents work, the more frequently users are pulled in for feedback — creating growing fatigue around "agent management."
To address this problem, Clone Labs is building a User Model: an AI that understands user intent and context without requiring constant intervention, enabling autonomous decision-making on the user's behalf. Rather than asking users to repeatedly explain context and step in manually, the User Model learns from users' behavior and preferences and communicates directly with AI agents.
The model is structured around three layers designed to maximize accuracy: Recording, which captures computer usage patterns; Memory, which accumulates and analyzes context and preferences during decision-making; and Prediction, which predicts and executes next actions based on those inputs. Clone Labs also applies a confidence-based gating system, automatically executing tasks when prediction confidence is high while escalating low-confidence or critical decisions back to the user. The company recently launched its first products, “Clone Desktop” and “Clone Plugin,” targeting AI-native builders who actively work with multiple AI agents.
Clone Labs was founded by undergraduate students from Seoul National University and has published seven research papers related to AI agents, including collaborations with Stanford University and Carnegie Mellon University. The team has also conducted research in key foundational technologies such as CUA (Computer-Use Agents), long-term memory systems, and privacy-preserving memory architectures — all essential components for building User Models.
Clone Labs was initially discovered through the 2025 NAVER D2SF Campus Startup Competition and received investment just three months after entering NAVER D2SF’s incubation program in January 2026. NAVER D2SF highly recognized the team’s rapid iteration across problem definition, hypothesis testing, and product validation within a short timeframe.
“As AI technology advances rapidly, entirely new bottlenecks and market opportunities are emerging alongside it,” said Sanghwan Yang, Head of NAVER D2SF. “Clone Labs quickly recognized that productivity is increasingly constrained by the bottleneck between humans and AI agents, and the team is tackling this challenge with strong research expertise and exceptional execution.” He added, “NAVER D2SF will continue to support founders who identify and experiment on the frontlines of the AI era, while expanding collaboration opportunities with NAVER.”
