Nexus: Startup Singularity Escape Raises Seed Round for an AI Platform for Team Agents

On August 2, 2026, Singularity Escape closed a seed round of tens of millions of yuan to develop Nexus, an AI operating system for team-based agent work. Stellar Capital and Shuimu Ventures led the round, with the accelerator Miracles Startup joining as a co-investor.

Why Solo Agents Fall Short in a Team

A typical scene in a modern team: the product manager hands requirements to one agent, the developer to a second, the marketer to a third. A few hours later, each agent has a finished result and the full context of its own task. But when the three pieces need to be merged, it turns out that nobody knows what the neighboring agent did, which decisions it made, or why.

• The round closed in August 2026; the amount is tens of millions of yuan
• Lead investors: Stellar Capital and Shuimu Ventures; co-investor: Miracles Startup
• Founder Xue Chuanyi holds a PhD from Tsinghua University and has publications at NeurIPS and CJA
• Product: Nexus, an OS for team collaboration with AI agents
• Key concept: a single graph of organizational state instead of isolated chats

Xue Chuanyi, CEO of Singularity Escape, calls this the "collaboration gap": humans become information intermediaries again, copying context between agents and explaining the same thing in different windows. The agent remains a "smart lone worker" rather than a full member of the team.

"Today, an agent is a capable individual, but it has not yet joined the organization," said

Xue Chuanyi, CEO of Singularity Escape, in an interview with 36Kr.

How Nexus Works

Nexus represents people, agents, tasks, knowledge and tools as interconnected nodes of a single graph. Each task carries more than a text instruction. It holds a full set of attributes: goal, owner, dependencies, access rights, current status, evidence of completion and acceptance criteria.

When a new agent joins a project, it sees the context of previous tasks, like a new employee who has received a complete onboarding. All results flow back into the shared graph and become the starting point for the next executor. This moves the agent from the category of personal tool into the category of organizational participant: one that can be delegated to, granted permissions, and replaced without losing context.

Nexus uses a two-level graph architecture. Inside the agent system, the nodes are goal, planning, specialized skills, memory and tools. At the organizational level, they are people, agents, tasks and knowledge. This structure makes it possible to pinpoint where an improvement occurred and which parts of the system it affects.

Self-Learning Under Evidence-Based Control

Self-learning in Nexus runs in three stages: real feedback from tasks → independent testing of candidate improvements → acceptance of changes through a governance mechanism.

The system can propose adjustments, such as changing how context is organized, how models are routed, or how workflows are configured. But no change goes straight into production. Each one first goes through independent testing and is compared on effectiveness, stability, cost and risk.

"Self-evolution is not unlimited autonomy.

It is stricter boundaries of evidence and governance," say Singularity Escape's official materials. The system must answer: why a change is needed, what exactly changed, whether it got better, and whether it can be rolled back.

According to Xue Chuanyi, Nexus's long-term value accumulates not in model parameters but in organizational state: task history, decisions made, and the evidence base. This asset does not reset when the underlying language model is updated.

What This Means

Singularity Escape is betting that business value is created not by a single powerful agent but by a team system that accumulates organizational intelligence. If the Nexus concept holds up in production conditions, the startup could carve out a niche as an infrastructure layer for companies that are already building multi-agent workflows.

Frequently Asked Questions

Who founded Singularity Escape?

Xue Chuanyi, a researcher from the post-2000 generation (born after 2000), holds a bachelor's degree and a PhD from Tsinghua University in reinforcement learning and multi-agent systems. He is first author of publications at the NeurIPS conference and in the CJA journal, and interned at SenseTime and Jizhi Robotics, where he developed algorithms for self-driving cars.

How large was the seed round?

According to 36Kr, the round totals "tens of millions of yuan" (千万级); the exact figure has not been disclosed. At the current exchange rate, that corresponds to roughly $1.4–10 million, depending on the exact amount.