Funding news · Sep 14, 2026
Xisiid Intelligence (羲悉智能) Raises ¥30M Seed for Brain-Inspired AI Agents
Xisiid Intelligence Raises ¥30M Seed Round
Xisiid Intelligence (羲悉智能), a Shanghai-based startup working on next-generation AI through brain-inspired intelligence, has raised a ¥30 million seed round. The funding was announced on September 14, 2026.
What Xisiid Intelligence Does
The company's core product is a self-evolving agent foundation system. It pairs two components:
- A lightweight proprietary model
- A hierarchical experience-memory network
These are organized in a dual System 1 / System 2 architecture — a design pattern inspired by how the brain handles fast, intuitive processing alongside slower, deliberate reasoning.
The goal of the architecture is to let AI accumulate and reuse task experience. Rather than treating each task as a fresh problem, the system is designed to build on prior experience so it can reliably deliver long-horizon professional work.
Xisiid Intelligence targets domains including:
- Finance
- Law
- Enterprise operations
These are fields where tasks tend to be lengthy, multi-step, and dependent on accumulated context — precisely the kind of work the company's experience-memory approach is intended to address.
Founding Team
Xisiid Intelligence was founded by Yansong Chua, a computational-neuroscience researcher. Chua previously served as a brain-inspired computing lead at A*STAR and at Huawei 2012 Labs.
The team spans multiple disciplines:
- Computational neuroscience
- Large-model algorithms
- Cognitive science
- Systems engineering
The blend of neuroscience and large-model expertise reflects the company's stated approach of drawing on brain-inspired principles rather than pursuing scale alone.
Round Details
| Field | Detail |
|---|---|
| Round type | Seed |
| Amount | ¥30,000,000 |
| Currency | CNY |
| Date | September 14, 2026 |
| Stage | Seed |
No lead investor or additional participating investors were disclosed with this announcement.
Why the Approach Is Notable
Most commercial AI agent development has focused on scaling large foundation models and layering tooling and orchestration on top. Xisiid Intelligence is taking a different route: a lightweight proprietary model combined with an explicit memory architecture.
That design choice carries several implications:
- Efficiency. A lightweight model paired with structured memory may reduce the compute required per task compared with relying on a very large model for every step.
- Experience reuse. A hierarchical experience-memory network is intended to let the system retain and apply what it has learned across tasks, which matters for long-horizon professional work.
- Dual-process reasoning. The System 1 / System 2 split mirrors a well-established cognitive-science framework, separating fast pattern-driven responses from slower deliberative reasoning.
What to Watch
With seed funding in place, the key questions for Xisiid Intelligence are how well its self-evolving agent system performs on real professional workflows in finance, law, and enterprise operations, and whether the experience-memory approach produces measurable reliability advantages over conventional agent architectures.
The company's Shanghai base and its founder's background at A*STAR and Huawei 2012 Labs place it within a broader Chinese research ecosystem that has invested heavily in brain-inspired computing.
Summary
- Company: Xisiid Intelligence (羲悉智能)
- Headquarters: Shanghai
- Founder: Yansong Chua
- Core product: Self-evolving agent foundation system
- Architecture: Lightweight proprietary model plus hierarchical experience-memory network, dual System 1 / System 2
- Target domains: Finance, law, enterprise operations
- Round: ¥30M Seed
- Announced: September 14, 2026