On May 5, Anthropic and Perplexity AI both unveiled new finance-focused AI products aimed at automating parts of the research, analysis, and operational work carried out across Wall Street and the broader financial industry.

At a launch event in New York on Tuesday, Anthropic introduced 10 AI agents designed for financial services firms including banks, insurers, and investment companies. The agents can perform tasks such as building pitchbooks, reviewing audits, drafting credit memos, conducting due diligence checks, and supporting financial analysis workflows.

The company also announced expanded integrations between its Claude AI platform and financial data providers including PitchBook, Moody’s, Morningstar, S&P Capital IQ, Dun & Bradstreet, and IBISWorld. Claude was additionally integrated more deeply into Microsoft Office tools such as Excel, PowerPoint, Outlook, and Word, which remain widely used across investment banking and financial operations.

Speaking at the event, Anthropic CEO Dario Amodei said financial services had become the company’s second-largest enterprise segment after technology. Yet, he also warned that software-as-a-service firms that fail to adapt to AI could “lose market value, go bankrupt, completely go bust.”

The announcements came as Anthropic continues expanding its enterprise business while reporting that its first-quarter revenue had grown “80 times on an annualized basis.” The company also had announced its $1.5 billion joint venture with firms including Blackstone, Goldman Sachs, and Hellman & Friedman to develop AI tools and services for private equity-backed companies.

Perplexity’s bet on research and analyst workflows

While Anthropic focused on institutional finance operations, Perplexity introduced a parallel push into financial research workflows. In a blog post, Perplexity announced “Computer for Professional Finance,” a finance-focused version of its Perplexity Computer platform. The company said the system supports more than 35 finance workflows and allows firms to connect licensed financial datasets from providers including PitchBook, Morningstar, Daloopa, and Carbon Arc.

According to the company, the platform is designed to automate multi-step research and analysis tasks rather than simply respond to prompts. Perplexity said the system can retrieve filings, compare financial data, synthesise analyst commentary, generate reports, and execute workflow chains across multiple tools and datasets.

“Teams with licensed data subscriptions can connect their own credentials through MCP connectors for Morningstar, PitchBook, Daloopa, and Carbon Arc. Teams without a terminal can use Computer's built-in finance tools, which draw on data from 14 providers, including Quartr and Fiscal,” the company noted in the post.

Although the two companies are approaching the market differently, both announcements reflect how AI firms are increasingly targeting specialised enterprise workflows rather than general-purpose chatbot use cases.

How the two approach finance using AI

Anthropic’s launch was centred on regulated financial operations such as compliance, underwriting, accounting, and due diligence. Perplexity’s focus is more research-oriented, aimed at analyst productivity and financial information synthesis.

For years, large parts of financial services have depended on analysts and associates to handle labour-intensive work including earnings summaries, industry research, diligence preparation, document review, and report generation. The latest generation of AI systems is increasingly being positioned as capable of automating parts of those workflows. This shows that AI is not a helper anymore, but a coworker, thereby leaning towards Agent-as-a-Service provision.

At the same time, executives across the industry have argued that AI is more likely to contribute to existing teams than fully replace finance professionals. JPMorgan CEO Jamie Dimon, who appeared alongside Amodei at the Anthropic event, has previously said the bank sees AI primarily as a productivity tool that can help employees work more efficiently.