A recent research report from Soochow Securities Company Limited indicates that the integration of AI and finance has advanced into the "intelligent agent" phase. In this new stage, the B2B sector is building competitive moats around the combination of data, content, and workflow, while the B2C sector is competing based on user access points, depth of scenario integration, and the closure of business loops. Financial technology companies possessing extensive financial data foundations, well-suited application scenario entry points, and mature experience in applying AI technology are expected to continue benefiting. Emerging online financial firms that demonstrate clear paths for business expansion and traffic monetization, substantial investment in AI research and development, a combination of user scale and quality, and sustainably high gross margins are poised to establish complete business ecosystems and maintain robust performance figures.

Where the competitive focus now lies

Since 2024, financial AI has evolved from handling isolated tasks like Q&A and summarization to the development of Agents capable of planning, using tools, and executing continuous operations. Starting in the second quarter of 2026, standardized interfaces like MCP and Skills have been deployed. This includes the launch of offerings such as Hithink RoyalFlush's iFinD MCP, Wind's AIFin Market, East Money's Miaoxiang Skills, Tongdaxin's MCP, and Hundsun Technologies' Juyuan MCP. This shift has moved financial data from gated terminals to standardized capabilities that Agents can call upon.

How vertical agents are penetrating financial scenarios

In the B2B market, several platforms are taking distinct approaches. Hithink RoyalFlush is integrating its fragmented AI functions, evolving from manual multi-tool use to an Agent that organizes capabilities. Wind is upgrading its Q&A assistance into an autonomous investment research Agent using its data infrastructure and institutional workflows. Meanwhile, Jinmen Finance, AlphaPai, and Gangtise are carving out differentiated paths based on roadshow content, AI-native products, and expert knowledge bases, respectively.

For the C2C market, Hithink RoyalFlush's Wencai has upgraded from natural language stock screening to an investment assistant with autonomous planning abilities. East Money's Miaoxiang has evolved from a standalone Q&A tool to an embedded AI portal within its app, connecting to its financial services ecosystem. Tongdaxin has introduced an "Agent Plaza" to transition from simple Q&A to a multi-Agent platform. Jiufang Zhitou Holdings is using an AI-powered stock machine as a hardware gateway that integrates investment advisory, research, and education services.

Two complementary paths for general Agents

General-purpose Agents are entering the financial sector through two complementary avenues. The first involves connecting to external professional financial capabilities. For instance, Alibaba's Tongyi Qianwen "Deep Research" now integrates data from over 13,000 stocks and roughly one million financial reports from Hithink RoyalFlush, while Moonshot AI's Kimi has connected to iFinD and Yahoo Finance. The second path is localized institutional deployment. Following the release of DeepSeek-V4 in April 2026, several leading brokerages, including Guotai Haitong, Industrial Securities, SDIC, Zhongtai, Sinolink, and Shanxi Securities, have completed their upgrade deployments. Hundsun's WarrenQ can now generate a 10,000-word investment research report in just five minutes.

Over the long term, behavioral data from the consumer side will feed back into model development, while professional data from the B2B side will strengthen vertical capabilities. This creates a data flywheel that drives continuous Agent improvement.

Why just four types of players matter

The competitive landscape is defined by four distinct categories of participants. Data owners like Hithink RoyalFlush, Wind, and East Money are competing on MCP openness, where data barriers set the floor and openness strategies determine the ceiling. Scenario owners, including Hithink RoyalFlush, Jiufang Zhitou Holdings, Fortune Trend, and Compass, are focusing on how deeply their scenarios are integrated to drive user stickiness, often using hardware as a differentiator. Model owners such as Doubao, DeepSeek, Tongyi, and Zhipu AI are finding that general models do not replace vertical ones but do absorb superficial C2C needs; DeepSeek, with its open-source and low-cost approach, has become the preferred choice for institutional deployment. Finally, platform owners like WorkBuddy represent both the largest customer and the biggest threat to data providers, as they have already integrated three financial MCPs.

Key risk factors to monitor

Potential risks include tighter regulatory oversight, slower-than-expected technological advancement, data security concerns, volatility in the equity markets, and delays in commercial adoption.