As many treasury teams start small with AI, one member company is taking a more ambitious approach, using Kyriba as a central hub to unify fragmented treasury systems and lay the groundwork for AI agents that can reconcile forecasts, explain variances and optimize cash decisions. - At a recent NeuGroup sponsored solution session, a senior treasury manager at a large technology company presented alongside Kyriba’s SVP of product solutions and strategy Thomas Gavaghan, who helped the member use Kyriba as part of a broader effort to build a treasury data lake and bring order to an environment that had been spread across multiple systems.
- "When we started this journey, we kept in mind that we are building a data foundation not just for the sake of just automated reporting—although that is great because everything is so manual—but it’s not end of the game," the member said. "The way we designed needed to be able to support future advanced analytics and AI initiatives."
A single source of truth. The treasury manager said the company started with “so many systems” and “no single source of truth,” with team members spending much of their time downloading files from different sources, manipulating them manually and then uploading elsewhere. The result was not just inefficiency but inconsistency: similar data could produce different numbers depending on which team was reporting it.
- The company was already using Kyriba as its treasury management system, but the broader treasury environment remained fragmented. Early efforts to improve reporting with tools like Tableau and Power BI exposed a deeper problem.
- “Every time we tried to do one dashboard, it turned into multiple projects,” she said, because not all of the underlying data was in the same place, and treasury needed new integrations and additional cleanup before a report could be trusted.
- That led the team to a broader redesign centered on building a data lake in Snowflake and connecting key systems to it over time. Kyriba was among the first priorities because, as the member put it, “day-to-day cash positioning, FX data, it’s all in Kyriba.” Mr. Gavaghan added that connectivity is critical because treasury operates across banks, ERPs, TMS platforms and other systems, making flexibility “essential as the world changes and companies adapt their architecture.”
Executive backing and dedicated resources. The project was a multiyear effort, with initial implementation taking three years, followed by additional dashboards and AI initiatives in the fourth. The company’s treasurer pushed the team to think beyond a short-term reporting fix and instead design a treasury data environment that could support the business for up to 10 years.
- “Without the executive sponsorship, we wouldn’t be able to prioritize as a company to build this,” the member said. Treasury was provided dedicated resources rather than relying only on shared IT support, including a treasury technology hire, a Snowflake developer and technical program management.
- The team planned about 50 use cases for the data lake, defined the inputs, processes and outputs involved and then prioritized the highest-impact systems and workflows. The goal was not merely to centralize information, but to create “consumption-ready” data that treasury could use immediately through dashboards and analysis tools.
The AI payoff. Once the architecture was in place, the company began using it for more than reporting. One early AI project was an “AI-powered free cash flow control tower,” designed to replace a spreadsheet-based process that had required gathering inputs from dozens of people each quarter and manually tracking whether results were on target.
- “Now we can put the model in Snowflake and then people can access information via Power BI or even a chatbot,” the member said. On top of that, the team layered in “AI agents to review the inputs, do forecast actual reconciliation and to layer on the machine learning model to do direct cash forecasting.”
- The long-term aim is broader still. She said the company is now thinking about “how do we use AI agents to perform what our capital markets team is doing,” including taking “a holistic view to take actions.” Mr. Gavaghan described that direction as moving from AI that assists treasury to AI that can increasingly “augment and replace parts of the work,” while still keeping humans in the loop for sensitive decisions.
