Big Tech Pushes AI Agents Deeper Into Finance

Summary

Major technology companies are moving general-purpose AI agents into finance, but they have not settled on whether the sector requires dedicated products. Tencent launched WorkBuddy Financial Edition on September 3 with more than 80 financial experts and expert groups for banks, brokerages, and insurers. Its workflows cover corporate-loan due diligence, fund research, and insurance customer operations; Tencent says WorkBuddy has entered more than 100 financial institutions since March, although brokerages contacted by the article said broad internal adoption has not yet occurred. ByteDance's Coze and Baidu's Kuku AI are currently more consumer-oriented: Coze exposes financial Skills for tasks such as financial comparison, ETF screening, and fund analysis, while Kuku AI combines financial data with retrieval, processing, and generation to produce research reports, presentations, and spreadsheet models. Kuku AI is described internally as a general office agent using finance as a demonstration of complex task execution, rather than as a separate financial product. The article identifies finance as attractive because high-value professionals still spend substantial time searching documents, checking data, updating models, and preparing materials, much of which is already digitized. It is also difficult because results must be accurate and traceable, permissions must be controlled, and regulated decisions may require human review and records of source data and reasoning paths. A June guidance document from China’s National Financial Regulatory Administration classifies areas including transactions, asset valuation, credit approval, insurance claims, and risk management as high-risk AI applications, while warning about data leakage, unauthorized access, tool misuse, and loss of control. Overseas approaches also differ: OpenAI and Anthropic extend general products through investment-banking plugins, templates, office integrations, and financial data sources, while Google launched a dedicated Gemini Enterprise for Financial Services with more than 50 financial Skills. Financial AI company Rogo, which reportedly serves more than 300 institutions and over 40,000 professionals, illustrates the market’s willingness to pay for connections between foundation models, specialist databases, source citations, and usable Excel, PowerPoint, and research outputs. The article concludes that financial agents may become a distinct software category if institutions pay for deeper workflow integration and governance, but stronger general agents, Skills, and connectors could limit the need for industry-specific versions.