When the Agent Reconstructs the Entrance: The Crossroads of Three Financial Information Service Providers
```
Agents are dismantling the terminal model that has been used in the financial information services industry for over twenty years.
In late May, Wind released the financial AI ecosystem platform AIFin Market, opening MCP capabilities to external Agents for the first time, enabling them to directly access Wind’s data, tools, and research capabilities.
Before Wind, the other two leading financial information service providers had already taken a step ahead:
On March 12, iFinD MCP was launched by Tonghuashun, aiming to let external Agents directly call its database; on the same day, East Money launched its Skills system, packaging information search, financial data query, and intelligent stock selection as callable modules.
On the surface, the three leading financial information service providers have reached a consensus on openness—extracting abilities once encapsulated in financial terminals and opening them up to Agents.
But what is truly noteworthy may not be MCP or Skill, but what was chosen first to be open, and what they tried to guard most.
For more than twenty years, financial information service providers have competed for terminal seats: fund managers open Wind, investors open Tonghuashun, stockholders browse East Money, data, information, and research tools are encapsulated in a single entry point;
As users begin to use Agents to query data, read announcements, select stocks, and analyze funds, data, research tools, analytical frameworks, and even investment methodologies are being decomposed from terminals and entering a new ecosystem in forms like MCP and Skill.
Today, the core question for financial information service providers has become: When users no longer directly open the terminal, at which link in the value chain can they still play a part?
Perhaps the three companies will eventually cover all links, but at the start of this transformation, what each chooses first to open and to protect still reflects their distinct genes.
Wind: The Terminal King Repositions
Although coming later in the timeline, Wind’s willingness to open its database to external Agents still surprised many.
For the past twenty years, Wind’s core moat has never been just data or even institutional clients themselves, but the workflow that institutional clients have accumulated inside the terminal;
Countless fund managers, brokerage analysts, and bank researchers rely on Wind for data, information, tools, and even to complete research work.
This terminal-winning model originated from US financial information giant Bloomberg.
In the 1990s, Bloomberg seized the opportunity as the global bond market expanded and mutual funds rose, evolving from a data tool to a work platform. By the early 2000s, many European and American investment banks included “proficiency in Bloomberg” as a recruitment requirement.
At the same time, Wind’s founder Lu Feng spotted the opportunity and successfully replicated Bloomberg’s terminal path in China.
This is why, regardless of recent technical advances, Wind continually focuses on keeping workflows inside the terminal.
In 2023, ChatGPT sparked the first wave of generative AI;
Bloomberg launched its own GPT with 50 billion parameters; Wind pushed the registration of “Wind Alice text generation algorithms” and gradually built an AI product matrix integrated within its terminal—intelligent customer service, investment advisor assistant, etc.
By 2025, as the Agent, multi-agent, and MCP ecosystems quickly develop;
Both Bloomberg and Wind continue to reinforce the terminal, with Wind releasing WindClaw, Alice Agent, and the Alice 27 smart financial OS, aiming to keep institutional workers within the terminal.
Yet, in this new technological wave, Wind’s terminal strategy faces significant challenges.
With rapid iteration of general models like ChatGPT and Claude, users’ expectations for Agents have increased: complex reasoning, open Q&A, cross-domain analysis, and long-text processing now matter greatly to user experience.
From compliance, adaptation, and autonomy perspectives, Wind’s AI products are based on self-developed large financial language models;
But Hub’s research found that Wind’s Agent performance in open analysis and complex reasoning is generally weaker than that of general models like Claude and GPT.
Some institutional users told Hub that the generated content sometimes contains data errors or lacks argument support; in complex task scenarios, toolchain stability needs improvement;
Hub’s testing showed that in tasks involving multi-layer data integration and in-depth analysis, Wind Agent occasionally responds with timeout or task interruption.
These issues don’t mean Wind Agent failed, but reveal a harsh reality: Wind is not good at foundational models, yet foundational models set the upper limit for Agent experience.
Meanwhile, competitors Tonghuashun and East Money have already opened MCP interfaces or Skill capabilities, integrating data and analysis modules into external Agent ecosystems;
As external OpenClaw, WorkBuddy, and HermeAgent can directly access databases and skill modules of other leading financial information providers, the workflow deposited in Wind’s terminal shows signs of loosening.
Possibly for this reason, Wind is trying for the first time to enter external Agents, so its capabilities stay in the new workflows.
In May, Wind released AIFin Market, opening its data, tools, and Skills system to external Agents, so users can access Wind’s capabilities via MCP without entering the terminal.
Hub’s testing found that external Agents can now instantly access data and perform tasks. The platform manages call frequency via a point system; free points cover basic scenarios like indicator queries and data extraction, more frequent or complex calls require additional point purchases.
However, errors in data aggregation still exist in queries with similar entity names.
Hub tested and found that some data for entities with the same name incorrectly accumulates; whether this needs structural optimization or further Agent ability improvements remains to be explored by the industry.
It’s worth noting that opening MCP does not mean Wind is abandoning the terminal.
As of June 5, Wind’s homepage core ad position has changed to its built-in smart financial OS, encouraging users to complete research work using Alice Agent, smart customer service, investment advisor assistant, and dozens of skills like mainline recognition, post-market review, and trading plans.
This shows that, for Wind at this stage, open Agent ecosystems and terminal reinforcement are parallel strategies, and AIFin Market adds new touchpoints besides the terminal;
The terminal can be decomposed, data can be called—but workflow still determines the pricing power of financial information providers. This may still be Wind’s main concern.
Tonghuashun: Answers Matter More Than Entry Points
Tonghuashun’s choices in the Agent era have formed a subtle mutual reference with Wind.
This March, Wind launched in-terminal AI products like Alice 27 and WindClaw, and opened MCP in late May; conversely, in March, Tonghuashun was first to launch iFinD MCP, hinting at a future self-developed Agent product, iFinD Claw.
In outcome, both companies are laying out terminal and open ecosystems at the same time, but the sequence differs;
Wind first reinforces the terminal, then opens its capabilities; Tonghuashun opens its data interface first.
Hub noticed that since launch in March, Tonghuashun has repeatedly expanded MCP coverage, stretching from financial databases to corporate business registration, risk, management information, etc. Now it supports stocks, funds, bonds, Hong Kong and US stocks, macroeconomic data, industry data, and announcement information.
Officially, iFinD MCP is described as “giving every lobster a professional financial database.”
This almost blatantly shows Tonghuashun’s ambition: no matter what Agent users employ—ChatGPT, Claude, or self-built by brokerages, funds, banks—they should be able to use Tonghuashun’s data for generating answers.
This choice is closely tied to Tonghuashun’s user base.
Wind serves institutional clients; Tonghuashun faces a massive population of individual investors.
By 2025, 57.43% of Tonghuashun’s revenue comes from advertising and internet marketing services, 32.35% from value-added telecom services, and only 3.6% from fund sales and trading services. Compared to traditional financial terminals, it's closer to an internet platform.
Different user structures create different product logic:
Wind cares about equipping terminals with more professional tools and accumulating workflow, whereas Tonghuashun’s concern is always how ordinary investors can get answers faster and better.
In 2013, Tonghuashun launched the smart financial search product “Wencai.”
For the first time, users needed no stock selection formulas or financial indicators—inputting in natural language let the system filter, query, and match. Wencai became one of Tonghuashun’s most representative products, driving rapid user growth during the 2015 bull market.
Tonghuashun’s product director Lu Zhongwen once said Wencai is the company’s core product, even “all resources are put into it.”
After ChatGPT’s explosion, Tonghuashun released the HithinkGPT financial large model, gradually integrating it into Wencai and iFinD terminals to further lower the threshold for financial information understanding.
Following this clue, Tonghuashun’s intent is consistent: shorten the distance between user and answer.
The technological boom has made new paths possible.
Since Agents can now actively call tools, retrieve info, and execute tasks—whether ChatGPT, Claude, Doubao, or brokerages, funds, and banks with internal Agents—they may become new work interfaces.
Before, users asked questions via Wencai; in the future, they may do so via Agent. The interaction changes—but reliance on data remains constant.
Lu Zhongwen notes that over the years, Tonghuashun has built iFinD database, natural language data retrieval systems, and various financial tools, all essentially tackling the same question: how to allow machines to accurately access financial data.
From this perspective, being the data provider behind Agents is far more important than competing for the Agent entry point itself.
Similar logic appears in overseas markets.
US financial data provider FactSet has long provided databases, analytical models, and open interfaces to brokers, funds, and asset management institutions. Entering the Agent era, FactSet quickly advanced its AI-ready Data and open interface systems, aiming for its database to enter more and more AI workflows.
Whatever Agent a user asks their question through, FactSet’s real concern is: when the answer is generated, is it using their data?
Tonghuashun follows the same logic.
Tonghuashun’s revenue is long built on huge user traffic, and the essence of traffic is demand. As long as they guard the database, demand will not disappear.
From this angle, opening iFinD MCP first is more like Tonghuashun’s step forward along its established path.
East Money: Guarding the Trading Side
Compared to Wind and Tonghuashun, East Money was first to open capabilities at a level more influential on cognition.
On March 12, East Money launched its Skills system for the Agent ecosystem, opening three abilities for external Agents: information search, financial data query, and intelligent stock picking.
Information search Skill can access news, announcements, research reports, and policy info;
Financial data Skill supports queries on stocks, funds, bonds, and company fundamentals;
Intelligent stock selection Skill can filter based on financial indicators, technical signals, and main business, among other conditions.
The open, well-packaged analytical capabilities reflect a more advanced goal—East Money seems more concerned about influencing investor decisions.
Exploring their business structure explains this choice better.
By 2025, East Money’s securities service revenue reached 12.535 billion yuan, 78% of total income; financial e-commerce service revenue was 3.182 billion yuan, near 20%; financial data service revenue was only 1.5%.
This means over 90% of East Money’s revenue comes from fund distribution and brokerage licenses that competitors lack.
In the late 2000s, East Money used its financial portal and stock forum to quickly gather a huge investor community;
In 2012, with the fund sales license, East Money for the first time bridged information and trading. Then Tiantian Fund rose in the internet fund sales boom, at one point monopolizing 80% of the third-party distribution market;
A year later, East Money acquired Tongxin Securities for 4.4 billion yuan, completing its transition from financial portal to wealth management platform.
Information attracts traffic, the stock forum influences cognition, Tiantian Fund handles fund sales, East Money Securities handles stock trading;
For decades, East Money’s core moat was a pathway that brought users right to the point of trading.
This is also Agent era’s biggest challenge.
Previously, investors used East Money to get information, form judgments, and trade. Now, more users are reading news, researching companies, and filtering funds via Agents.
Wind is challenged on workflow, Tonghuashun on the data entry point, while East Money is challenged in their ability to connect cognition and trading.
This may be why East Money focuses on Skills.
Agents help users get information but don’t naturally possess mature investment frameworks and research systems;
East Money has encapsulated its abilities in information interpretation, stock analysis, and fund research as Skills—not to provide more info, but to keep its research capabilities involved in user decision making.
Similar explorations can be seen overseas.
For example, US independent research provider Morningstar has advanced the integration of generative AI with its research system, embedding fund ratings, portfolio analysis, research reports, and investment insights into AI workflows, and launching AI assistant tools for advisory and institutional clients.
Such institutions want their accumulated research frameworks and analytical abilities deeply embedded in investment decision making.
From financial portal to Tiantian Fund, from East Money Securities to Miaoxiang Skills, every East Money transformation in the past two decades was essentially about moving closer to trading.
Wind guards workflows, Tonghuashun guards data, but East Money cares most whether, when trading happens, it still stays in that user path.
Risk Disclaimer and WaiverThe market has risks, and investment needs caution. This article does not constitute personal investment advice, nor does it take into account the special investment objectives, financial situation, or needs of individual users. Users should consider whether the opinions, views, or conclusions in this article suit their specific circumstances. Investment based on this is at your own risk. ```