From the 2026 Technology Finance Symposium – West
For finance teams, the real opportunity with AI isn’t simply having better conversations with a large language model. In a discussion moderated by Daniel Olsen, principal at Baker Tilly, panelists KC Fike, director at Baker Tilly, and Sowmya Ranganathan, co-founder and CEO of Lumera, explored the bigger opportunity: embedding AI into business processes, where it can understand context, follow rules, and take on work that otherwise consumes employees’ time.
That means looking beyond isolated prompts and prototypes. Fike and Ranganathan emphasized that an effective AI agent needs to understand the process it’s operating within: where its data comes from, which policies apply, what outputs are expected, and when a human needs to make the final decision. Context is particularly important when policies and business information change over time.
Contract review provided a practical example. Ranganathan demonstrated how AI can perform the initial work of comparing a sales contract against company policies, identifying provisions that require attention, and directing issues to the appropriate finance or legal reviewer. Instead of producing a lengthy summary that creates another task for employees, the technology can flag specific problems and suggest changes directly within the workflow.
Human accountability, however, remains central. If a contract requires both finance and legal approval, the system should enforce those responsibilities rather than allowing an unauthorized person, or an AI agent, to make the decision. Permissions, audit trails, and clear distinctions between what an agent can read, write, or approve become important controls.
Successful deployment also depends on meeting employees where they already work. The panel noted that even an impressive AI application can become an abandoned side project if employees have to visit a separate dashboard. Integrating agents with tools such as Slack or Teams and existing systems of record makes AI part of the workflow rather than another destination.



