AI company Anthropic, valued at $900 billion, has launched 10 specialized agents that are impacting the job security of financial professionals on Wall Street and posing challenges to data service providers. In May, during a closed-door financial services event on Wall Street in New York, Anthropic CEO Dario Amodei officially introduced 10 financial-specific AI agents. These agents cover the entire workflow chain, including investment banking, research reports, private equity, wealth management, and fund operations, comprehensively restructuring Wall Street workflows from front-office investment banking to back-office fund management.

Market reactions have been mixed. Following the announcement, shares of U.S. financial data service providers plummeted, with FactSet dropping over 8% intraday and Morningstar falling nearly 3%. In contrast, bank stocks remained stable; Goldman Sachs, JPMorgan Chase, and Bank of America did not decline. The market perceives AI as a tool rather than a threat, believing it can enhance profit margins.

The core market concern is the extent of the impact these 10 agents will have on financial professionals. Essentially, these agents are not designed to replace jobs but to handle 80% of the manual labor, allowing financial professionals to focus on high-value judgments, client communication, strategic decision-making, risk assessment, and innovative business development. From an international perspective, only those engaged in repetitive tasks face genuine risk. Some securities firm executives maintain an open and collaborative attitude towards AI technological innovation. Wang Hongtao, Chief Information Officer of Sinolink Securities, stated, "How you use AI determines how AI will empower you. You give it purpose; it illuminates your path." AI is evolving from a "back-office tool" to an "AI digital colleague" working alongside employees. The core concept behind the first batch of 18 digital employees launched by CITIC Securities aligns with this perspective. Yu Xinli, the company's Chief Information Officer and head of the Information Technology Center, explicitly stated that digital employees will inevitably reshape the financial landscape. However, the ultimate goal is to enhance the work experience for all employees and drive the securities industry towards breakthroughs in both personnel capability enhancement and revenue growth.

It is evident that AI technological innovation is profoundly impacting both domestic and international financial markets. How will Anthropic's 10 agents affect financial operations and professionals? What insights do they offer for the AI progress of domestic securities firms and funds?

Question 1: What can the 10 AI agents do? The 10 agents released by Anthropic are end-to-end workflow automation systems designed for specific financial business scenarios. Functionally, they cover research and client services, fund operations, and finance, with clear divisions of labor:

1. Pitch Agent: Generates fundraising pitch decks by integrating comparable company data (comps), leveraged buyout models (LBO), and other information to quickly produce branded presentation materials, assisting investment banking teams in preparing for client meetings or fundraising proposals. 2. Meeting Prep Agent: Automatically compiles client background information, market dynamics, and counterparty details before client meetings, generating briefing packages to provide structured references and save team preparation time. 3. Market Researcher: Based on industry keywords or targets, gathers information from multiple data sources to generate industry overviews, competitive landscape analyses, peer comparisons, and investment idea lists, assisting researchers or investors with market research. 4. Earnings Reviewer: Analyzes earnings call transcripts or financial report documents, extracts key information in real-time, updates financial models, and generates draft research notes, helping analysts quickly understand changes in financial reports and their potential impact. 5. Model Builder: Supports the construction of various financial models, including discounted cash flow (DCF) models, leveraged buyout (LBO) models, and integrated three-statement models, providing automation from data input to model setup to facilitate valuation and forecasting by analysts. 6. Valuation Reviewer: Reads valuation packages provided by general partners (GPs), runs valuation templates, performs cross-verification, and generates limited partner (LP) reports to ensure the reasonableness and compliance of valuation results. 7. GL Reconciler: Handles the reconciliation of general ledger (GL) accounts with subsidiary ledgers, automatically identifies discrepancies, traces root causes, and generates approval materials, assisting finance teams in completing reconciliations and audit preparations. 8. Month-End Closer: Executes month-end closing processes, including accruals, roll-forwards, and variance explanations, generating closing reports to ensure the accuracy and completeness of financial data. 9. Statement Auditor: Conducts audit checks on LP statements before distribution, verifying consistency, completeness, and audit readiness to ensure reports meet compliance requirements. 10. KYC Screener: Parses new client account opening documents, checks compliance via a rules engine, flags missing or anomalous items, and generates compliance reports, assisting compliance teams in completing client due diligence (KYC) processes.

These 10 agents do not operate in isolation. Anthropic simultaneously released office software plugins, enabling tasks initiated in spreadsheets to seamlessly transition to presentation software for report generation while maintaining contextual consistency. Furthermore, through model context protocols, these agents are already integrated with 11 major financial data providers, including Moody's, Dun & Bradstreet, Morningstar, S&P Global, and FactSet. This means a typical investment banking analyst workflow can be restructured: using natural language instructions to invoke an agent, automatically pulling data from multiple sources, generating analytical models in a spreadsheet, and then automatically transferring to presentation software to create materials, with the entire process maintaining an audit trail.

According to Anthropic, its underlying large model scored 64.37% on financial agent benchmark tests, ranking first in the current industry.

Question 2: What is the real impact on financial professionals? The release of these 10 agents has elicited complex reactions among financial professionals. On one hand, the allure of efficiency gains is evident. On the other, the perennial question of "Will AI replace analysts?" has gained renewed urgency. According to sources within foreign investment banks, recruitment for junior analyst positions is experiencing a downturn, with many graduates from top-tier universities unable to secure internships in such roles. Anthropic's intervention most profoundly affects these junior analysts, whose roles involve significant time spent on manual data analysis, slide creation, and other tedious, repetitive tasks—work that AI can now replace. This trend suggests a more certain decline in demand for junior research positions. When research team leaders and senior analysts can leverage intelligent research assistants for foundational work, the need for junior analysts naturally diminishes.

Conversely, the importance of senior chief analysts and institutional sales roles may increase. The former possess irreplaceable value due to their professional knowledge and industry experience in determining key data sources and assigning data weights. The latter's value in driving business development and maintaining client relationships may become even more pronounced in the AI era. Industry insiders note that an analyst's core competitiveness traditionally lay in data organization, information synthesis, and basic modeling. AI now efficiently empowers these aspects, requiring analysts to focus more on high-value-added tasks such as logical reasoning, causal analysis, market forecasting, and strategy formulation. Simultaneously, cross-disciplinary comprehensive abilities, such as understanding AI technology and constructing interdisciplinary analytical frameworks, will become crucial future competencies for analysts. The goal of leveraging agents is to liberate analysts from "information processing," allowing greater participation in value judgment and distribution.

A more forward-looking perspective comes from Huang Yanming, Director of the Orient Securities Research Institute. In a public interview, he divided research business into four steps: collecting information, processing information, creating ideas, and disseminating ideas. He noted that AI can significantly enhance efficiency in the first two steps, but its role is limited in achieving human-level thinking and innovation based on vast information.

Question 3: Overseas traditional financial data service providers face impact; what advantages do domestic counterparts have? The decline in U.S. financial data provider stock prices reflects the impact of Anthropic on their businesses. In contrast, what domestic data providers similar to FactSet and Morningstar exist, and how can they reinforce their competitive moats?

Wind Information is perhaps the closest domestic equivalent to a "Chinese Bloomberg," nearly monopolizing the financial terminal market for domestic institutional investors. Its moat stems from years of accumulated data coverage and consistency, long-established user habits among institutional clients, and an almost irreplaceable professional standing in industry standards like bond valuation and futures data. Wind's Alice is continuously being upgraded and optimized. The newly launched Agent Swarm Mode on May 15 transforms it into a schedulable financial AI project team, enabling AI to function like a project group.

Tonghuashun's advantage lies in its retail investor traffic and data accumulation in certain scenarios, but it has yet to form a complete layout in the institutional business agentization direction. East Money Information transitioned from a financial portal website to a financial data terminal and then obtained a securities brokerage license. Its core strength lies in monetizing the retail investor traffic market.

Companies like financial IT service provider Hundsun Technologies, while not a data company, have long served the trading systems of securities firms and funds. Their advantage lies in proximity to business processes; their various engineering capabilities represent precisely the most challenging aspects for agents to penetrate core financial workflows.

From a policy environment perspective, the operating landscape for domestic financial data companies differs fundamentally from their overseas counterparts. On one hand, data sovereignty and security compliance requirements strictly limit the business scope of overseas giants, forcing domestic institutions to choose almost exclusively among local suppliers when procuring data services. This provides a natural "protective cushion" for domestic companies. On the other hand, domestic companies also face compliance hurdles such as algorithm filing and data security assessments when advancing agent products. These are constraints that can be addressed through investment but also constitute real barriers to entry.

Question 4: How will the 10 agents affect domestic financial institutions? Changes in the data company landscape are merely superficial; a deeper impact lies in how agents are altering the competitive logic of financial services. Examining the impact of Anthropic's 10 agents on domestic securities firms and the fund industry requires consideration from multiple dimensions.

Technological paths differ, each with its own advantages. Observations indicate that leading domestic securities firms are not lagging in AI deployment. Huatai Securities formally established its "All in AI" strategy for 2025, launching an AI-native financial trading terminal and building an AI-driven intelligent research system. Guotai Junan Securities and Haitong Securities jointly released the trillion-parameter multimodal securities vertical large model "Junhong Lingxi," whose AI customer service has replaced over 60% of human consultations. China Merchants Securities proposed the "AI Securities Company" strategic vision, aiming for 100% AI coverage in customer service, business scenarios, employee operations, and high-frequency tasks by 2030. East Money Information, relying on its self-developed "Miaoxiang" large model, launched the "AI Researcher" agent capable of autonomously completing information gathering and analysis processes. Institutions like China Securities and CITIC Construction Securities have also successively launched intelligent research report processing workbenches.

However, the differing paths of the two systems determine their respective strengths and weaknesses. Anthropic's 10 agents enable financial institutions to combine different agents like building blocks, rapidly integrating multiple data sources into a single workflow. The advantage of this model lies in its high degree of standardization and fast deployment speed, making it suitable for mature process-oriented overseas investment banking businesses. Domestic securities firms have taken a different path in AI deployment, focusing on deep customization around local market needs. Fundamental differences exist between A-share information disclosure rules, regulatory approval processes, investor structures, and those of overseas markets, making simple replication of overseas templates difficult to implement effectively. Domestically developed intelligent investment banking platforms and AI documentation systems for investment banking by Chinese securities firms may hold advantages in aligning with local regulatory requirements.

From a data ecosystem perspective, Anthropic's 10 agents are directly integrated with 11 major financial data providers. Analysts can seamlessly invoke data from different sources within the same workflow, with every conclusion traceable to its original data source. While domestic securities firms are also building data middle platforms, challenges such as fragmented data procurement, technical hurdles in cross-source data integration, and data silos may persist. Additionally, compliance issues facing agents in the securities industry cannot be ignored. For example, in investment banking, regulations clearly define responsibility for the authenticity and accuracy of issuance materials. Underwriters and signing sponsors bear personal legal liability for documents. How is content generated by agents legally defined? Must signatories assume full responsibility for agent-generated content? This legal vacuum in responsibility delineation may represent an institutional barrier for agents entering core investment banking processes.

The impact on the fund industry may differ from that on securities firms. In the public fund sector, AI application concentrates on quantitative investment and intelligent investment advisory. Among Anthropic's released agents, those most relevant to the fund industry are Valuation Reviewer, GL Reconciler, and Month-End Closer. These tools targeting fund operation back-ends address the pain point of high operational costs in public fund middle and back offices. If agents can compress a significant portion of back-office operational expenses, it will substantially impact the fund industry's business model.

In terms of penetration paths, due to high compliance requirements and strict risk control systems, the adoption speed of agents in public funds may be slower than in securities firms. Leading quantitative private equity institutions incorporated machine learning into factor mining and strategy construction processes years ago, with their grasp of AI technology potentially exceeding that of some securities firm research institutes. Since these institutions have already built AI systems more tailored to their own strategies, the incremental value of Anthropic's financial agents may be relatively limited for them.

Beyond the impact on individual business points, a more significant concern is the fundamental challenge agents pose to the sell-side research business model. For a long time, commission sharing has been the core revenue source for securities firm research institutes. If agents can replace a significant portion of sell-side research services—such as automatically completing earnings commentary, industry data tracking, and basic modeling—the direction of public funds' willingness to pay for sell-side research becomes a question worth pondering.

From Anthropic's practice, it is clear that the core development trend of current financial AI agents lies in deeply integrating with institutions' existing tools and ecosystems, and connecting data to the entire business process. For the domestic securities and fund industry, this presents both pressure and opportunity. The pressure stems from the accelerating pace at which leading overseas financial institutions are embracing AI. The opportunity lies in the fact that localized domestic capital market scenarios and regulatory requirement adaptations may represent advantages difficult for overseas competitors to replicate.