The age of fragmented spreadsheets is over. Embat's Theo Wasserberg discusses how agentic AI is dissolving the 30% automation barrier in treasury, shifting the treasurer’s focus from manual reconciliation to strategic execution, and delivering 90%+ forecasting accuracy that redefines the function as a core business driver.
The corporate treasury function stands at a critical inflection point. Amidst relentless market volatility and the escalating complexity of global operations, the mandate for real-time, strategic insights has never been more urgent. To move beyond endless manual reconciliation and truly harness their data, treasury leaders are increasingly turning to advanced AI.
We spoke with Theo Wasserberg of Embat, an AI-native treasury management platform, to understand how this technology is fundamentally transforming the treasury landscape and elevating the role of the finance professional.
The primary challenge facing corporate treasurers today is the deeply manual and fragmented nature of their operations. As Theo Wasserberg notes, the work is often a patchwork of disconnected efforts:
“It’s not looking at one spreadsheet. It’s looking at 15 different spreadsheets that are all interlinked, with data in different places. You have to keep going to different bank accounts to pull a CSV file and then try to consolidate.”
Traditional Treasury Management Systems (TMS) and ERPs, built on rigid, rule-based logic (“if X, then Y”), inherently struggle with real-world complexities—such as customers combining invoice payments or entering incorrect references. This creates a hard ceiling where only about 30% of tasks can realistically be automated.
Embat’s integrated solution is engineered to remove this ceiling. Leveraging an API-first architecture that connects to over 15,000 financial institutions globally and provides native, bidirectional ERP integration (including SAP, Oracle NetSuite, and Microsoft Dynamics 365). The system allows treasurers to express intent, and the AI autonomously manages exceptions, achieving up to 99% automation and delivering true, real-time centralization.
The shift to AI isn’t simply about efficiency; it’s about strategic empowerment. True treasury transformation means moving the function from merely a support role to a strategic driver of business performance. Treasury moves from “drifting off to the side of finance” to being core to business strategy.
For finance leaders, measuring the ROI of advanced AI requires looking at both time-to-value and strategic impact. Where legacy systems can take 12–24 months just to implement, AI-native solutions must show measurable results rapidly.
Key performance indicators (KPIs) that benchmark this strategic uplift include:
The core innovation Embat brings to the table is ‘TellMe,’ an agentic AI analyst tool. For the treasury professional, “agentic” means the system can autonomously pursue objectives without needing step-by-step instructions. Treasurers simply state the goal, and the system intelligently executes the plan.
This is a significant step beyond simpler automation. Generic Large Language Models (LLMs) often require users to invent use cases, while TellMe is built around predefined, finance-specific functions where outcomes are measurable and auditable. Its multi-layered architecture—featuring a Perception Layer for continuous data ingestion and a Cognitive (Reasoning) Layer trained on data from over 250 corporate finance teams—interprets complex scenarios and plans multi-step actions.
The real value of this approach lies in its prescriptive power—the ability to move from passive reporting to proactive decision-making. A powerful example lies in stress-testing liquidity.
If a liquidity threshold is breached, the AI autonomously:
In a sector defined by the need for accuracy and auditability, trust is non-negotiable. TellMe is engineered for integrity through data containment and transparency. It maintains a tamper-proof, time-stamped audit trail for every AI action, enabling instant verification for auditors and ensuring continuous compliance. This is governed by a human-in-the-loop model, where users retain control over validation and approval, while the AI performs repeatable, high-confidence actions.
The ultimate outcome is the fundamental elevation of the treasurer’s role. Theo Wasserberg sees AI as augmentation, not replacement:
“You don’t automate someone’s job; you automate the painful, manual parts and augment the human user.”
By automating data work, treasurers can pivot from managing exceptions to strategic execution, focusing on high-value functions like M&A support, capital structure optimization, and liquidity strategy. This makes finance careers more attractive to new talent, who “don’t want to spend time processing data, you want to type your intent into a system and get answers”.
Looking forward, Theo predicts that the next major disruption won’t be a new technology, but improvements in open banking connectivity – standardized and persistent regulations that unlock another wave of innovation.
Ultimately, the perception of treasury will change: it will be seen not as a complex niche, but as an essential, strategic, data-driven function.