PayPal Tracks $53 Billion in Treasury Cash With an AI Agent

PayPal AI

Highlights

PayPal deployed a conversational AI agent in treasury operations covering $53 billion in funds, allowing employees to analyze cash positions, currency exposure and financial risks through natural-language queries.

PayPal’s broader treasury overhaul eliminated 90 spreadsheets, saved about 1,700 hours and increased cash visibility from 78% to nearly 100% across roughly 1,000 accounts.

Generative AI use for cash flow management rose from 16% to 80% of finance teams, PYMNTS Intelligence found, but only 5% use agentic AI for cash flow management.

Ask PayPal’s treasury team where its cash sits, and the answer now comes from a chat window. A Deutsche Bank case study describes how that change unfolded inside the company. The system handles $53 billion in corporate cash and customer funds, and the team once relied on 90 spreadsheets to track it. Cash visibility across about 1,000 accounts now runs through the same interface.

    Get the Full Story

    Complete the form to unlock this article and enjoy unlimited free access to all PYMNTS content — no additional logins required.

    Subscribe to our daily newsletter, PYMNTS Today.

    By completing this form, you agree to receive marketing communications from PYMNTS and to the sharing of your information with our sponsor, if applicable, in accordance with our Privacy Policy and Terms and Conditions.

    PayPal’s treasury operation now runs on one system with a conversational agent on top, according to the Deutsche Bank case study. Can Balcioglu, PayPal VP and treasurer, whom the case study quotes on the rollout, says the approach “will define how treasury functions in the future.” The study credits the system with coverage across 160 currency pairs and cash visibility that rose from 78% to nearly 100% across about 1,000 accounts. It also reports accuracy above 99% in cash positioning and about 1,700 hours saved.

    PayPal has not disclosed how much of that time becomes cost savings. Its April 2026 newsroom release says thousands of employees can query company data in plain language, which suggests the approach extends beyond treasury. The case study says two companion agents, for counterparty data and bank-account data, are still in development.

    Centalion’s Agent Prepares Transactions for Staff Approval

    Centalion, a large independent commodities trader, uses an agent to prepare trade finance transactions, and treasury staff decide whether to proceed. Centalion’s agentic treasury release describes a connector that links the agent to Komgo’s Konsole platform, a FinTech system focused on trade finance. Staff ask in plain language whether there is enough facility capacity to issue or amend an instrument, and the agent prepares the transaction.

    The release says the first deployment, on a syndicated facility, cut issuance processing time by about 55% and amendment processing time by about 80%. It also projects more than 60 working days of treasury capacity freed each year if the agent extends across the company’s full trade finance activity.

    J.P. Morgan’s agentic cash management release describes an agentic workflow for overnight cash work that flagged a $2.3 million currency mismatch. Centalion’s release says governance and approval checks run in the system’s orchestration layer, which keeps the decision with staff while the agent handles preparation.

    Generative AI Use for Cash Flow Reaches 80%

    PYMNTS Intelligence’s 2026 Working Capital Index, “Working Capital Has Lost Its Strategic Edge,” found that 97% of the lowest-performing middle market firms use AI in treasury, nearly matching the 100% rate among top performers. The index treats seeing cash and controlling it as separate capabilities.

    PYMNTS Intelligence’s Time to Cash report, produced with J.P. Morgan, found that generative AI use for cash flow management rose from 16% to 80% of finance teams. Nearly two-thirds, 64%, now use AI-generated insights when making cash flow decisions. Those figures measure AI used for insights, not autonomous action. Agentic AI use for cash flow management stands at 5% of finance teams.

    For treasury leaders, the question is less whether an agent can answer a cash query than who signs off when it prepares a move.

    For all PYMNTS AI coverage, subscribe to the daily AI Newsletter.