“The quality of the questions you ask an LLM is directly related
to the quality of the output - and that’s a skill finance teams have to build.”
in-class processes in reusable agent templates. Over time, a treasury process perfected in one industry, such as oil and gas or pharmaceuticals, can be ex- posed, securely and selectively, to peers in other industries, effectively turning operational excellence into a shared, agent-driven asset class. Use AI To Lower Risk, Not Just Lift Productivity Much of the public AI narrative centers on efficiency, but for finance and treas- ury leaders the more compelling story is risk. Under volatile macroeconomic con- ditions, the ability to view and evaluate liquidity, exposures, and counterparty risk in near real time becomes a strategic differentiator. Kyriba’s TAI is designed to bring gov- erned intelligence and automation to liquidity management: agents assist with cash-flow forecasting, anomaly detection, reconciliation, and scenario analysis, all within controlled treasury environments. By keeping models and data inside the Kyriba platform, the company reduces the attack surface and addresses the concern that highly pro- prietary liquidity data might be repur- posed or leaked by external providers. For CFOs, that translates into a more consistent, explainable view of risk, AI- augmented, but still anchored in existing governance structures and sign-off pro- cesses. For IT and security leaders, the embedded-LLM design and emphasis on
Tom Callway
observability support regulatory expec- tations around data residency, privacy, and model risk management. Callway summarizes the philosophy simply, “Data integrity, human oversight, and transparency are the foundational pil- lars that transform AI from a promising technology into a trusted business tool for finance and treasury operations.” What Leaders Should Do Next The story that emerges from Kyriba’s journey is not about a single product; it is about how finance, treasury, and IT leaders navigate an inflection point where AI becomes critical infrastruc- ture. For CFOs and treasurers, the most urgent task is to clarify where AI should lower risk and increase resilience, li- quidity visibility, anomaly detection, and reconciliations before chasing ge- neric productivity gains. They must also define what “trust” means in the context of their own governance and control re- quirements. From there, they can spon- sor a move beyond pilots by backing a small set of high-value, agentic use cases with clear human oversight and insist on transparency into how models use and transform financial data. For treasury and finance leaders lead- ing teams across the office of the CFO, the priority is to turn AI from an ab-
stract innovation theme into concrete workflow change. That means working with IT to identify where prebuilt agents can safely take friction out of daily pro- cesses, codifying best-in-class practices into reusable agents, and building team skills in prompt engineering and critical review so staff can interrogate AI outputs rather than accept them at face value. For CTOs, CIOs, and their direct re- ports, the advice centers on architecture and governance. Prioritize platforms that embed models securely inside ex- isting data boundaries, support full au- ditability of agent behavior, and align commercial terms, often via consump- tion-based models, with realistic adop- tion curves. At the same time, design for an ecosystem future in which treasury and finance agents must interoperate with AI capabilities in ERP, CRM, and other enterprise platforms through se- cure, standardized protocols. Across the entire organization, the com- mon action is to treat trusted, human-cen- tric AI as a strategic capability and to lead its adoption deliberately. The inflection point Callway describes is already here, agentic AI is moving from demos to daily use, and finance and treasury leaders who delay, risk being outpaced by peers who can see, decide, and act faster with trusted AI at their side.
let’s connect sapinsider.org 45
Made with FlippingBook - Share PDF online