While many dealmakers remain reluctant to adopt agentic artificial intelligence (AI) tools, the technology will inevitably transform M&A practices, experts say.

Agentic AI refers to machine-learning technology capable of acting autonomously to achieve specific goals in unpredictable environments, making it a potentially revolutionary technology.

“M&A practitioners still have some privacy and compliance concerns about the tech,” according to James Lehnhoff, Datasite’s chief technology officer.

However, dealmakers also had similar concerns with previous innovations, including virtual data rooms (VDRs) themselves, Lehnhoff said. VDRs from companies like Datasite have replaced physical data rooms over the last 25 years.

M&A practitioners will continue to show caution around data and intellectual property (IP) when it comes to new tech, but “AI adoption is inevitable,” Lehnhoff said.

Agentic AI can be seen as a sliding scale, with self-driving cars as the most complex example, Lehnhoff said. “At the simpler end of the scale, agentic AI can combine several more basic tasks, such as categorising content, summarising it, naming it.”

Deal practitioners began to use AI seriously for tasks like data extraction about four or five years ago, according to Ken Bisconti, senior vice president and co-general manager of SS&C Intralinks, which also provides VDRs. Large language models (LLMs) are also beginning to be used within M&A platforms, he added.

“Our customers, including bankers, hesitate to use these tools with sensitive information,” Bisconti said, adding that all AI within M&A needs to respect user permissions for security and traceability.

Paying attention to prompt engineering and iterative processes based on feedback will let M&A firms bring their proprietary methodologies into automated workflows, Bisconti said.

Over the longer term, agentic AI should be seen as “the future of financial services,” according to Model ML co-founder and CEO Chaz Englander.

Model ML, which has offices in New York, London and Hong Kong, has been working on developing “super agents,” which can handle complex tasks like generating earnings summaries. It is interested in acquiring engineering-led companies with task-specific AI agents and five to 20 team members, its CEO said earlier this month.

Luminance is another innovator. The London-based company is developing tech that will allow AI systems on either side of contract negotiations to handle routine talks and identify risks, CEO Eleanor Lightbody told Mergermarket in March.

Practitioners seek comfort

At the moment, M&A professionals are looking at the reasoning behind decisions as they gain comfort with the tech, according to Prakash Kanchinadam, Head of AI and Analytics at SS&C Intralinks. “Humans are part of the decision-making process at this stage,” he said.

Dealmakers who want to start testing agentic AI took should seek to stress the underlying data while adding human judgement calls into the mix, according to Nadine Mirchandani, EY-Parthenon deputy global vice chair.

“Users need to challenge results, add a human element, add uncertainty scenarios in option planning, and test different outcome scenarios,” Mirchandani said. At the same time, M&A practitioners need to think of agentic AI as “an assistant rather than a replacement,” she said.

Although agentic AI will be assistive, Lehnhoff said that the initial risks involve it giving incorrect results with confidence.

One way of thinking about the implementation of new tools is to focus on effectiveness and not just efficiency, Mirchandani said.

Indeed, EY-Parthenon’s research shows that deal teams that emphasise human elements of transactions – “nurturing talent, bringing cultures together thoughtfully, and thinking about missions” – get better results, Mirchandani said.

Routine processes in line for disruption

Commonplace processes are first in line for disruption. “M&A professionals are focused on getting comfortable using AI agents for routine, repetitive and time-consuming tasks,” Bisconti said.

Meanwhile, Lehnhoff said: “Agentic AI is coming for anything with complex workflows.” He added that the tech should be seen as part of a wider trend, which involves “drudgery getting abstracted away.”

Agentic AI can be used systematically to speed up M&A processes, Mirchandi said, adding that AI-based financial analysis is improving exponentially.

However, the approach of assistive agents also carries risks given that junior practitioners typically honed their craft with routine tasks at the beginning of their careers.

The winning strategy in this context is to reinvent workflows, according to Matt Beane, assistant professor in the Technology Management Program at the University of California, Santa Barbara, and author of The Skill Code.

“What is dealmaking if agents are getting 4x to10x more capable and less expensive per year?” Beane asked. “With an answer here, new roles will emerge – new bundles of tasks that make sense – and the main thing is to optimise collaborative interaction there.”

The transition will be messy, Beane said, adding that reinventing processes to suit the new tech will be the key. “You don’t want to optimise for yesterday,” he said.

Use cases

Junior bankers who organise VDRs will be able to use agentic AI, Kanchinadam said. “The nightmare scenario involves putting the wrong document in the wrong place – AI can help reduce the risk while also increasing the speed,” he said.

The tech can be used to offer a pipeline of possible targets, according to Charis Stengos, public sector account executive at Google Cloud, which operates Agentspace. Agents will be able to advise on strategy as well, he said.

As technology trust levels increase within M&A practices, use cases will move from insights to autonomous and agentic actions, Kanchinadam agreed.

One use case, for example, is identifying pain points in due diligence, such as similar questions being asked with different phrasings, Kanchinadam said.

Agentic AI can also help with post-deal integration. Some private equity (PE) firms have begun experimenting with new data and analytics tools that can give them an integrated overview of all the assets in a portfolio, even if they have idiosyncratic systems. JMAN is a company that offers bespoke data-driven services for PEs, for example.

The new tech enables the possibility of adding an agentic layer on top of legacy systems instead of migrating them, Mirchandani said.

Although M&A practitioners are cautious about taking the first steps to incorporate agentic AI into their workflows, there are no doubts that the time will come.

The end game of the coming revolution in M&A will be “data-driven decision-making with a human element,” Mirchandani said.