Artificial intelligence agents — tools designed to take on workplace tasks — could fundamentally reshape the corporate finance function in coming years, rendering some jobs obsolete while creating new ones, according to Georg Glantschnig, vice president of Dynamics 365 AI ERP at Microsoft.

The technology is still in the early stages of development, and some analysts say it may be too soon for rapid adoption when it comes to functions such as corporate finance.

Eventually, however, AI agents could act as “skilled virtual coworkers, working with humans in a seamless and natural manner,” according to a McKinsey report published last year.

Microsoft envisions every business process in an organization being impacted by AI agents, Glantschnig told CFO Dive. Such agents can help finance teams automate labor intensive processes like the financial close, he said.

The following is a Q&A, conducted via email, between Glantschnig and CFO Dive’s Alexei Alexis. The exchange has been edited for clarity and brevity. 

CFO Dive: How do you define AI agents or agentic AI, and how much of a potential game-changer is this emerging space when it comes to business automation, particularly in the corporate finance function?

Georg Glantschnig: An agent uses AI to execute tasks and take action on your behalf. Agents work alongside or on behalf of a person, team, or organization and range from simple, prompt-and-response agents that carry out a simple query or a single task like invoice approval to more advanced, fully autonomous agents that orchestrate end to end processes such as order to cash.

Ensuring accuracy and compliance of financial operations and reporting is a fundamental responsibility of every finance department. With financial data dispersed across multiple data sources, financial transactions flowing in and recorded from internal and external sources, maintaining financial data accuracy is a daunting task, often done manually and prone to human errors.

By leveraging agents to automate key financial processes, finance professionals can benefit from time saved searching across systems without sacrificing, or even increasing, accuracy and compliance. 

Furthermore, the agent’s ability to dynamically adapt and constantly learn makes an agent-driven automation far more sustainable than a traditional rule-based one that requires more ongoing maintenance.

CFO Dive: Are fully autonomous AI agents a reality or just an aspiration at this stage and is the market ready for this?

Georg Glantschnig: Our customers’ trust and confidence in AI is top of mind for us. Therefore, we started with assistive capabilities in 2024 across our full ERP portfolio. Today, 50% of our users are using Copilots in Dynamics 365 apps. In the same responsible way we’re introducing agents, gradually increasing their autonomy at a pace suitable for each individual customer. Different customers and teams differ in their AI journey and how quickly they adopt new technologies, but we already seeing many examples for customers moving quickly.  

Customers are already building their own agents with Microsoft Copilot Studio. For example, Pets at Home, the UK’s biggest pet care company, created an agent for its profit protection team to more efficiently compile cases for skilled human review, which could have the potential to drive a seven-figure annual savings. Organizations like McKinsey & Company, Thompson Reuters, Clifford Chance, and more around the world are also seeing incredible impact from building their own custom agents in Copilot Studio.

Copilot for Finance also provides both an assistive and automated agent experience to complete data reconciliation, enabling the user to choose the level of automation required. With assistive, users are involved in each step of the process and can “graduate” to the fully agentic experience once confident.

CFO Dive: From your company’s perspective, how important is the human element when it comes to corporate finance adoption and use of AI agents, and how might that change as the technology evolves and becomes more mature?

Georg Glantschnig: Agents automate business processes and can vary in complexity, scope, and autonomy — from generating insights on request, to running discrete tasks, to dynamically planning, managing, and orchestrating complete processes independently. Agents won’t interact with end users directly but with those users’ copilots. In AI-first companies, we anticipate that every employee will have their own copilot, and they will be supported by tens of thousands of agents. Our objective is to have a copilot for every worker and an agent for every business process.

We anticipate that with time, humans will take a more supervisory role, overseeing the work done by the agents, and the agent’s learning processes. 

Agents created in Copilot Studio are limited by the actions and tools that are explicitly defined by the author. Each interaction an author has with an agent is logged, so authors are able to review the record to see which actions are called by the agent and why.   

Agents created for Dynamics 365 include “human in the loop” approvals. For example, the Account Reconciliation Agent in Dynamics 365 Finance automatically reviews and reconciles transactions between subledgers and the general ledger. When it finds discrepancies, it escalates them to a human for further review.

Humans remain in control by selecting the functions the agent handles and can update these settings at any time. The agent suggests solutions for discrepancies based on previous human resolutions. As it works alongside humans, the agent learns and proactively proposes updates to rules and actions enabling it to gradually become more autonomous. At any time, humans can review the agent's operations, ensuring full visibility and establishing trust.

CFO Dive: Is there a risk that certain roles within the corporate finance function could be displaced by AI agents in the future? If so, how much of this could we see? 

Georg Glantschnig: AI agents are helping revolutionize many financial workflows, by automating and streamlining labor intensive processes like financial close. By doing so, financial professionals are able to spend more time focusing on providing strategic insights and driving business growth, while ensuring accuracy and organizational compliance. We are seeing AI agents help drive similar shifts across sales, service, and other business functions and are optimistic about the future of AI-first business processes.

We see the AI revolution as one that is fundamentally reshaping work and how work gets done. While it’s too soon to determine the exact impact on the finance workforce, it’s likely that some jobs will disappear, and some new ones will emerge. As mentioned above, administrative jobs might turn into supervisory roles, and we’re seeing a new role of “agent business administrators” emerging — with a responsibility for setting up and continuously improving the agents.