I firms have been fighting for enterprise attention and dollars. Now, they’ve got a new target: Finance.Finance pros become AI's next target

Several major AI firms have launched new tool initiatives aimed at finance professionals. Seeking to embed themselves and scale within a particularly lucrative market, these tools come as both OpenAI and Anthropic aim to boost revenue and race towards profitability ahead of their public offerings.

Here’s what was announced:

• Claude financial service agents: Anthropic released ready-to-run agent templates for financial services tasks, including building pitch books, screening KYC files, reviewing valuations and closing books. Claude also now works across Microsoft Excel, PowerPoint, Word and soon Outlook through the Claude add-ins for Microsoft 365.
• OpenAI and PwC’s Native Finance Function: The ChatGPT maker and professional services firm announced a collaboration to build agents around the “core operating rhythms” of finance, including planning, forecasting, payments, treasury, taxes and accounting. The organizations noted that the partnership is already in progress, building a procurement agent inside OpenAI’s finance organization.
• Perplexity Computer for Financial Services: The self-described AI answer engine has extended the reach of its “general-purpose digital worker” to finance. Now, finance teams can bring licensed data from providers like Morningstar and Pitchbook into the agent to work with 35 dedicated workflows for tedious analyst work.

While the upside is obvious for AI firms, these tools also present real potential to democratize financial analysis, Terra Higginson, a principal research director at Info-Tech Research Group, told The Deep View. Until recently, AI has been “notoriously bad at financial analysis.” But with access to proper data, these tools have improved vastly, she said. Now, private investors could have access to tools that would have previously only been available to the top tier of investors and institutions.

“I still would not blindly use any model for financial analysis … but this is the first time the direction feels usable as a tool and not just a huge risk,” Higginson said.

However, risks still exist. Along with the obvious data security and regulatory risks, if everyone is using the same models, leveraging the same data, and coming to the same conclusions, “markets could become more crowded, more reactive, and potentially more volatile,” Higginson said. The real edge isn’t in trusting outputs outright, she said, but rather sharpening human judgment and knowing when not to trust the model.

“We always need to keep a human in the loop. I cannot emphasize this enough,” Higginson said. “Finance is highly regulated, high-stakes, and way too risky to let models operate alone.”

Our Deeper View

The problem with implementing AI into any risky industry is always speed. Now more than ever, executives are feeling the pressure to get returns from their models, with many CEOs feeling their job is at stake if they don’t get it right as soon as possible. That pressure could easily lend itself faster, and potentially sloppier, deployments. In financial contexts, that sloppiness could have dire domino effects. So while these tools could completely transform the world of finance, enterprises, investors and professionals may want to tread carefully.