AI Foundation Software Deployment and Application in the Financial Industry in the Era of Large Models

Introduction: The talk focuses on the deployment and application of AI foundation software in the financial industry during the era of large models, sharing experiences from a bank.

Part 1 – Application trends: AI technology has evolved from machine learning to ModelOps, enabling OCR, multimodal data, and AIGC, which are reshaping marketing, risk control, and operations in finance, improving efficiency and reducing costs.

Part 2 – Design thinking: In the large‑model era, AI foundation software should adopt a new architecture comprising AIaaS and data‑driven pipelines, knowledge‑graph management, explainability tools, and automated tuning, supported by vector databases and a full ModelOps stack.

Part 3 – Deployment outlook: Large models will enhance front‑office chatbots, middle‑office risk reports, and back‑office code generation, while challenges include engineering complexity, reliability, and cost‑effectiveness; a layered platform with compute, storage, and model services is proposed.

Conclusion: Combining large and small models will enable agile, intelligent financial services, creating a unified super‑application ecosystem.