Agentic artificial intelligence is set to transform corporate treasury functions by automating up to 80% of routine coordination and processing tasks. According to a new report released by EY India, this shift would allow treasury professionals to pivot away from low-value administrative work and focus on high-impact strategic decisions, such as risk management and capital allocation. The findings suggest that while the technology offers significant efficiency gains, successful implementation depends on establishing robust data foundations and governed workflows before scaling.
Currently, treasury teams at mid-levels spend approximately 60% to 70% of their bandwidth on manual, low-value activities. These tasks include gathering data from disparate systems, performing manual validations, and preparing routine reports. The reliance on manual batch reconciliation remains a widespread challenge, with over 50% of global corporations still depending on these inefficient processes. This reliance on manual intervention not only consumes valuable time but also introduces operational risks and limits the agility of treasury departments.
The Shift Toward Intelligent Treasury Models
The EY report highlights that a mature treasury operating model, supported by agentic AI, can fundamentally change how liquidity is managed. By automating routine coordination, the remaining 20% of the treasury function can be dedicated to activities requiring human judgment, such as strategic funding and complex risk assessment. The objective of this transformation is not to reduce headcount, but to reallocate existing talent toward activities that generate greater business value.
Beyond simple automation, the integration of AI agents is expected to significantly enhance liquidity forecasting. The report indicates that forecast accuracy of approximately 90% is achievable for 30-, 60-, and 90-day liquidity periods. Improved accuracy in these forecasts allows companies to reduce excess liquidity buffers, thereby freeing up capital that would otherwise remain idle. This shift represents a move from reactive treasury management to a more predictive, intelligent operating model.
Data Architecture as the Foundation for Success
EY cautioned that companies must address fragmented data systems before attempting to deploy agentic AI at scale. The report emphasizes that the path forward requires a centralized treasury data lake that integrates information from enterprise resource planning (ERP) systems, banking platforms, contracts, emails, and market data. Without this single source of truth, AI-based decision-making cannot be considered reliable or secure.
“The path forward runs through the datalake, not around it,” the report stated. Organizations are advised to prioritize the creation of a Treasury Center of Excellence (CoE) to oversee data pipelines, workflow libraries, and governance frameworks. This centralized oversight is essential for ensuring that AI agents operate within defined parameters and that data integrity is maintained across the entire treasury ecosystem.
A Phased Approach to AI Adoption
For organizations looking to begin their transformation, EY recommends an “advisory before autonomous” approach. This strategy involves using AI to provide recommendations that are reviewed and approved by human professionals before any action is taken. The report specifically advises companies to avoid autonomous payment execution during the initial pilot phases to mitigate risk and build confidence in the system.
Companies are encouraged to start with smaller, lower-risk use cases such as cash forecasting, cash reconciliation, and handling exceptions for Know Your Customer (KYC) and Anti-Money Laundering (AML) processes. Reconciliation and reporting are highlighted as particularly effective starting points because they involve high volumes of manual effort and offer clear, measurable outcomes. As treasury functions become increasingly data-driven, the ability to build governed workflows will determine which organizations successfully realize the benefits of agentic AI and build more resilient operations.
