2024 China AI Agent Industry Report: Definitions, Architecture, Market Size and Trends

Definition and Core Architecture

An AI Agent (Artificial Intelligence Agent) is an intelligent entity capable of perceiving its environment, making decisions, and executing actions. Unlike traditional AI that relies on prompt‑based interaction, an AI Agent can independently think, invoke tools, and achieve a given goal. A typical LLM‑based AI Agent system consists of four components: the Large Language Model (LLM) as the brain, Memory, Planning, and Tool usage.

Human‑AI Collaboration Modes

Three collaboration patterns are identified:

Embedding mode : AI capabilities are embedded within applications.

Copilot mode : AI acts as an assistant, providing real‑time suggestions.

Agent mode : An autonomous agent interacts directly with users and the environment, offering the highest efficiency and expected to become the dominant future model.

Features and Classification

AI Agents exhibit autonomy, interactivity, reactivity, and adaptability. They are classified into two major types:

Autonomous Agent : Operates independently with self‑learning capabilities.

Generative Agent : Generates content or actions based on generative AI techniques.

Market Size

Global autonomous agent market revenue is projected to grow from US$345 million in 2019 to US$2.929 billion by 2024, indicating rapid adoption and the emergence of software‑assistant forms.

Ecosystem and Vendor Landscape

The Chinese AI Agent market is in its early stage, with diverse participants such as AIGC‑native firms, internet giants, enterprise SaaS providers, RPA vendors, low‑code/no‑code platforms, and 3C hardware manufacturers, each leveraging domain advantages to enter the market.

Key ecosystem diagrams illustrate the relationships among platform layers, vendors, and application scenarios.

Business Models

Common monetization approaches include:

Software and services

Agent‑as‑a‑Service (AaaS)

LLM‑as‑a‑Service

Agent Store

Consumer services

Enterprise solutions

On‑demand platforms

Data and analytics

Technology licensing

Development Trends and Agent Types

Academic research classifies agents into four categories:

Logic Agent : Processes input language/multimodal data and generates output.

Task Agent : Decomposes specific tasks, plans and executes actions without long‑term memory.

Job Agent : Handles abstract responsibilities, maintains environmental perception and memory, and self‑generates sub‑goals.

Self‑evolving Agent : Pursues autonomous learning and evolution, encompassing most technical challenges.

Currently, most commercial products focus on Task Agents due to their technical maturity and ease of replication. In the short term, Job Agents are expected to accelerate development and gradually evolve toward self‑evolving agents.

Future Outlook

Job Agents will gain momentum, and the industry will continue moving toward agents with autonomous learning capabilities, reshaping human‑machine interaction across countless sectors.