How FIS is shaping AI-Driven business processes
When discussing the use of AI in the energy industry, people often focus on specific use cases, weighing benefits against challenges related to trust, security, and efficiency. But why not let customers decide where AI should be applied—and where it shouldn’t? This is precisely what FIS aims to offer.
The company envisions an AI-based framework that can be configured to automate various non-standard business processes. Harshad Kolpyakwar, Head of Energy Product Management at FIS, recently explained to me how this approach will work.
At the core of this concept are AI agents—specialized tools trained to perform specific tasks. These agents can be orchestrated into workflows that require some degree of automation but cannot be fully automated. Processes that follow predefined steps to achieve a result are straightforward to automate and do not require AI intervention. However, more complex processes, such as, for example, issue resolution, typically involve numerous manual steps that vary depending on the situation. These processes are often labor-intensive, combining both predefined steps and unique, case-specific actions. To optimize such workflows, intelligent tools are needed—alongside human oversight to approve or refine AI-generated results.
AI agents in this framework handle smaller tasks, such as searching the web for specific information, retrieving market prices or other market data, and extracting key details from unstructured files, etc. The orchestration framework then integrates these agents to support a customer’s specific workflow. The final result is presented to the user for approval or adjustments, with user feedback helping the AI agents learn and avoid repeating mistakes. “The human must always be in the loop for such non-standard workflows,” Harshad emphasized.
Rather than predicting where customers might want to apply AI, FIS plans to develop several demos for some use cases and present them to potential users. This approach encourages customers to identify the most valuable applications for their business, while the framework allows for rapid configuration of these customer use cases.
All necessary components for this solution are already available. Companies developing large language models (LLMs) and AI tools—such as ChatGPT-4 and Gemini—provide the AI agents, while orchestration frameworks are accessible as both open-source and licensed software. According to Harshad, FIS is now finalizing the proof-of-concept (PoC) phase, focusing on selecting the orchestration framework and finishing demonstration use cases. With this stage nearly complete, the next steps will be guided by customer demand.
Keep in touch and sign up to our Newsletter
