How Technology and AI are Redefining Investment Banking

Think of the M&A deal process as the journey driving home through a city at rush hour. AI agents are your navigation system that helps reduce the complexity of your journey. However, for the navigation to really provide support, it requires the appropriate roads, rules, and driver to successfully complete the journey. This Article explores AI’s direct impacts and necessary foundations in the journey as companies adopt their own AI technology through one key question: do AI agents reduce or increase M&A deal complexity? The first section focuses on investigating areas in which AI leads to a reduction in deal complexity, while the second one shifts the attention to points of increased intricacy.

Complexity in M&A deals is shifting, and the net benefit of implementation of AI depends on whether the foundations of businesses are set to handle new technologies. AI agents do not magically simplify deal flows, automate work, and create outputs, they reallocate complexity. They raise the bar required when it comes to data architecture and governance. The question the industry should ask is not “AI or not”, but where to lean on agents, and what foundations must be in place to avoid failures in this transition.

Where complexity drops

Advisors and investors who choose to adopt AI are likely going to experience a boost in efficiency in various key areas: pre-deal analytics, due-diligence processes, and securing post-deal value. Through the implementation of AI agents, parties will leverage agent capabilities that allow them to improve the M&A process for firms.

• Pre-deal analytics (identifying acquisition targets)

Through the implementation of AI agentic innovation, advisors can leverage pre-deal and mid-deal analytics. There are three keyways in which AI can be leveraged:

Market scanning and watchlists

Scanner agents run 24/7. They watch fillings, earnings calls, news, governance changes, and track hiring signals. When these factors shift, they track it into a queue summarizing key changes, why they matter, and who they are going to impact. The implementation of AI in market scanning enables a more informed and consistent control over the market.

Dynamic peer construction and benchmarking

Building an accurate comparable set of companies to benchmark a client’s company is crucial in the M&A process. Matcher agents construct dynamic peer sets from what companies do: products, customers, geographies, growth, and margin patterns. As a peer’s profile evolves, the set refreshes automatically improving day-to-day comparisons of employees and the quality of information to serve clients.

Pitch and valuation pack

Looking beyond market and competition analytics, valuation and narrative is a central focal point in the M&A process, and AI can help in drawing conclusions to be inserted in the final pack. The goal is not a finished book summarizing information; it is a robust draft so teams can collaboratively discuss assumptions and strategy moving forward.

In practice, the largest players are already implementing this technology into the M&A process. UBS’s in-house M&A co-pilot scans a universe of roughly 300,000 companies in less than half a minute, identifying buy-side ideas and potential buyers in sell-side situations (Doenecke, 2024)[1]. This tool emphasizes the power of AI that already exists in deal flows. While the firm noted limitations in current capabilities of AI on other deal stream, what can we expect AI’s capabilities to be in three years from now?

There is a reduction of complexity in search, similarity, and synthesis. This results in a tighter M&A deal loop with a broader coverage of information and fewer dead ends. Such innovation enables employees, specifically analysts, to dive deeper into the most significant issues and tasks. They use synthesized shortlists to personalize outreach and test valuation assumptions. As a result, deal makers spend less time asking for materials and more time strategizing with clients. Overall, there is less tracking and more thinking, so firms can focus on what matters most.

• Document-heavy due diligence

This is where the implementation of AI agents presents the biggest win. On the sell side, AI’s benefits come from collecting information and uploading it into virtual data rooms (VDRs), structuring folders, and scrubbing sensitive information. On the buy side, AI comes into play through reviewing large amounts of contracts, financials, HR and compliance documents. It is in repetitive, high-volume, and coordination-heavy tasks where AI agents help most. Concretely, the implementation of AI reduces the need for employees to scan through piles of documents. This enables employees to focus on more strategic value creations as hours, or even days, are cut from data organization and project management tasks. Nearly four in five companies applying generative AI report lower manual effort, freeing up time to focus on judgement rather than paperwork (Jeff Haxer et al., 2025)[2]. In an M&A case on an internal reorganization, a law firm reviewed 12,000 documents three times faster, reducing client costs by 90% (Dickson, n.d.)[3].

Efficiency at this level allows firms to take on larger tasks and deliver on previously unrealistic timelines. Beyond speed, accessibility lowers complexity in the process: agents can pull internal and public data content into one searchable workspace with source links and timestamps, standardize outputs, and instantaneously identify missing documents from reports. Moreover, AI agents strip out the complexity out of the process by automating high-volume tasks and synthesizing information that allows for a quicker and more informed process. As you move to “securing long-term post-deal value,” this same AI backbone becomes the operating layer for synergy realization, issue escalation, and ongoing performance monitoring.

• Securing long-term post-deal value

After AI strips complexity from due diligence, it shifts its value into integration. For larger deals, post-deal integrations can take from 9-24 months, which presents a plethora of opportunities for growth, but also opportunities for failure (Hofmeister, 2025)[4]. AI agents turn that into a major opportunity by connecting systems to accelerate synergies and create a better operating model. When looking into how AI agents can secure long-term value, we are talking about turning a newly formed business into a more durable, future-looking business. This entails aligning the operating model, capturing cost and revenue synergies, and key performance indicators (KPIs). Here AI agents do not stand as an external perspective, they are embedded as the operating layer. Agents pull information across functional departments and systems to plan workflows and monitor performance indicators so issues surface early and integration strategies hold. Accenture’s PMI research shows that teams embedding agentic AI across the deal lifecycle are far more likely to capture post-acquisition value, with benefits in achieving speed, securing synergy delivery, and enabling “leapfrog innovation” to the operating model. Looking at real-world applications of this, Alvarez & Marshall’s “A&M Assist” utilizes advanced predictive algorithms to provide early warning sights and critical insights in the post-deal phase (Cortes, 2025)[5]. By actively observing operational performance, the tool can pinpoint hurdles and preserve value. As an example, if employee attrition rates unexpectedly rise in departments, AI system will identify this trend and propose solutions. Through early action, clients can rapidly address challenges and continue driving value gains.

While presenting clear advantages, the adoption of post-deal AI systems is not as common compared to other stages of the process. According to an Accenture’sstudy, 82% of organizations have integrated GenAI into pre-deal M&A workflows, yet post‑deal integration remains an underused opportunity, implying strong potential for differentiated gains (Accenture, 2025)[6]. While these agentic systems bring speed and control to integrations, they also enable employees to narrow their focus. Instead of tracking performance across various fronts, agents leverage signals into a prioritized queue of issues to address. Each issue is recorded with its context, impact, and next steps, so teams can dive deeper, solve root causes, and drive sustainable outcomes that bring long-term value.

The opportunity for AI adoption is clear, but it is not a plug-and-play solution. To maximize value, companies face a maze that requires companies to adopt systems, change skills and work culture, and redesign processes. That is where complexity rises, and where we turn to next.

Where complexity rises

Companies that adopt AI agents are likely going to experience various improvements in the quality and speed of output; however, the advantages of AI are only reaped if the right foundations are in place. The hard part today is in facing increasingly larger hurdles: a modern data backbone, confidentiality, people risk, and addressing biases that AI could look passed. Companies that successfully address these hurdles will propel the business forward, turning speed into trust and creating a more resilient operating model that scales into the next deal.

• Modern data backbone

AI enables various business innovations; however, the most important feature that it requires to function is data. For firms to maximize output, or even implement AI into their operations, they require a modern data backbone that will facilitate the collection, storage, and movement of data.

Agentic implementation does not fix messy data; rather, it amplifies it. With clean, governed,

time-sensitive data, agents can reduce complexity in all the ways previously listed. However, without the necessary infrastructure, AI projects risk being abandoned by firms. Deloitte’s M&A research is clear that multi-agent systems only deliver when built on modern data architecture (Dragon, 2025)[7].

Building the backbone is a multi-faceted project and large data programs often take time and capital. Research concluded by McKinsey and Oxford found that big IT projects run 45% over budget and 7% over time on average. BCG’s C-suite survey similarly reports delays and budget overruns (Palumbo et al., 2024)[8].

These various reports display the foundational challenges of implementing AI. For firms that do take on the challenge, it is all about accepting a shift of attention to data structure from stages and time spent on the M&A process.

• Confidentiality and environment

Creating the appropriate infrastructure for data is a challenge, but it does not stop there. Firms must also foster a safe and secure environment for data privacy, and they achieve this through clear data classification, information barriers between teams, role-based access, and more. EY frames its data infrastructure as “AI enabled and safeguarded”, emphasizing protected workspaces for sensitive deal data. Agents must live within the walls. There must be a clear line on what agents can leverage for their outputs. However, tight controls can reduce data access to agents, which limits its space to maximize utility. EY reports stress that AI must be seamlessly embedded into each procedure, but giving agents freedom can cause a significant discussion on privacy, potentially fostering a sense of untrustworthiness to stakeholders. This must be carefully addressed by firms, and it is where it can become tricky because a single mishandled dataset can trigger a series of penalties and destroy trust. The focus must be to build the environment first, then agents can safely operate at deal speed.

• People risk and change management

The implementation of AI agents is not just in the technical aspect of handling data and infrastructure but managing the people side. Mercer’s research places the people risk as the focal value lever to address across deal lifecycle. Agents only deliver if people adapt to the workflows. To maximize the output of agents, the organization must foster openness to technological adoption. Today, there is a gap in skills and confidence, with BCG finding that only 6% of companies have begun upskilling in a meaningful way, despite AI being a top priority. In this change to leverage AI advantages, firms must prepare their people to continue driving value side-by-side with AI. The industry cannot solely depend on agents to achieve output. Firms that pair agent rollouts with clear roles and real training for their people will see quicker implementation and steadier integrations. People first, then agents. Without the proper training and change management, technological adoption will not stick.

• Dealing with biases

AI agents can mirror human decisions in seconds. However, they also keep the biases we can have in our decisions. Professor Steven Nason, professor at the Hong Kong University of Science and Technology, shares how we tend to see the stock prices of deals in an M&A fall post merger, which led him to evaluate how AI could play a role in the M&A process to see if deals should or should not be done. He mentions how, when asking AI for advice on how to proceed with a deal, it will generally give a textbook response on factors that should be considered and a list of tasks that should be conducted to answer our question. On the contrary, when providing AI with all the historical information and asking if, in each deal, prospect theory is present, it identifies the bias and says the deal should not be done. Prospect theory is a psychological framework that emphasizes how when people are faced with losses, they become more risk seeking while, when faced with gains, they become more risk averse. This highlights how AI can behave just like humans. Therefore, for firms that decide to integrate AI into their systems, it is crucial that the proper training and data is provided so that biases are addressed in the output.

Going back to the journey we began this Article with: if AI agents are the navigation system, then they have proven that they can cut through traffic to reduce complexity. But the safe arrival still requires the basics. Firms that invest in AI and secure the fundamentals can turn speed into reliability and trust. The route is set; now firms must ensure they have the essentials in place to reach their destination safely.

Authorship Disclosure:

Head of the Learning and Research Division: Lorenzo Ferrante

Learning and Research Division Analysts: Emmanuel Puno

Bibliography 

1. Doenecke, P. (2024, September 10). UBS Has an AI Tool That Can Scan 300,000 Firms in 20 Seconds. Bloomberg.com; Bloomberg. https://www.bloomberg.com/news/articles/2024-09-10/ubs-has-an-ai-tool-that-can- scan-300-000-firms-in-20-seconds
2. Jeff Haxer, Maja Omanovic, Ben Siegal, & Brooke Houston. (2025, February 4). Generative AI in M&A: You’re Not Behind—Yet. Bain. https://www.bain.com/insights/generative-ai-m-and-a-report-2025/
3. Dickson, H. (n.d.). AI-POWERED UNDERSTANDING OF KEY CONTRACTUAL FEATURES AI for the Spectrum of Contract Review Projects FLAGS ANOMALIES AND RISKS 1,000+ LEGAL CONCEPTS AUTOMATICALLY DETECTED AI FOR ALL CONTRACT REVIEWS TEACH AI NEW CONCEPTS VIA POINT-AND-CLICK TRUSTED BY 500+ ORGANIZATIONS. Retrieved November 27, 2025, from https://www.theedgeroom.com/wp-content/uploads/2023/01/Diligence-Product-Sh eet.pdf
4. Hofmeister, R. (2025, November 7). Unleash the Power of Agentic AI in Post-merger Integration. Accenture.com; Accenture. https://www.accenture.com/us-en/blogs/business-functions-blog/agentic-post-merg er-integration
5. Cortes, M. (2025, April 9). Using A&M’s AI agent to improve deal value and transform M&A execution (A. Kumar, Ed.). Alvarez and Marshall. https://www.alvarezandmarsal.com/insights/using-ams-ai-agent-improve-deal-value-and-transform-ma-execution
6. Accenture. (2025, September 2). Unleash the power of agentic AI in post-merger integration. Accenture. https://www.accenture.com/us-en/blogs/business-functions-blog/agentic-post-merger-integration
7. Dragon, L. (2025, July 22). Where is the value of AI in M&A: why multi-agent systems needs modern data architecture. Deloitte. https://www.deloitte.com/cz-sk/en/services/consulting/blogs/where-is-the-value-of- AI-in-MA-why-multi-agent-systems-needs-modern-data-architecture.html
8. Palumbo, S., Rehberg, B., & Li, H. (2024, April 30). Software Projects Don’t Have to Be Late, Costly, and Irrelevant. BCG Global. https://www.bcg.com/publications/2024/software-projects-dont-have-to-be-late-cos tly-and-irrelevant?utm_source=chatgpt.com
9. Tax, I. (2025). How AI will impact due diligence in M&A transactions. Ey.com. https://www.ey.com/en_ch/insights/strategy-transactions/how-ai-will-impact-due-di ligence-in-m-and-a-transactions?utm_source=chatgpt.com

---

[1] Doenecke, P. (2024, September 10). UBS Has an AI Tool That Can Scan 300,000 Firms in 20 Seconds. Bloomberg.com; Bloomberg. https://www.bloomberg.com/news/articles/2024-09-10/ubs-has-an-ai-tool-that-can- scan-300-000-firms-in-20-seconds

[2] Jeff Haxer, Maja Omanovic, Ben Siegal, & Brooke Houston. (2025, February 4). Generative AI in M&A: You’re Not Behind—Yet. Bain. https://www.bain.com/insights/generative-ai-m-and-a-report-2025/

[3] Dickson, H. (n.d.). AI-POWERED UNDERSTANDING OF KEY CONTRACTUAL FEATURES AI for the Spectrum of Contract Review Projects FLAGS ANOMALIES AND RISKS 1,000+ LEGAL CONCEPTS AUTOMATICALLY DETECTED AI FOR ALL CONTRACT REVIEWS TEACH AI NEW CONCEPTS VIA POINT-AND-CLICK TRUSTED BY 500+ ORGANIZATIONS. Retrieved November 27, 2025, from https://www.theedgeroom.com/wp-content/uploads/2023/01/Diligence-Product-Sh eet.pdf

[4] Hofmeister, R. (2025, November 7). Unleash the Power of Agentic AI in Post-merger Integration. Accenture.com; Accenture. https://www.accenture.com/us-en/blogs/business-functions-blog/agentic-post-merg er-integration

[5] Cortes, M. (2025, April 9). Using A&M’s AI agent to improve deal value and transform M&A execution (A. Kumar, Ed.). Alvarez and Marshall. https://www.alvarezandmarsal.com/insights/using-ams-ai-agent-improve-deal-value-and-transform-ma-execution

[6] Accenture. (2025, September 2). Unleash the power of agentic AI in post-merger integration. Accenture. https://www.accenture.com/us-en/blogs/business-functions-blog/agentic-post-merger-integration

[7] Dragon, L. (2025, July 22). Where is the value of AI in M&A: why multi-agent systems needs modern data architecture. Deloitte. https://www.deloitte.com/cz-sk/en/services/consulting/blogs/where-is-the-value-of- AI-in-MA-why-multi-agent-systems-needs-modern-data-architecture.html

[8] Palumbo, S., Rehberg, B., & Li, H. (2024, April 30). Software Projects Don’t Have to Be Late, Costly, and Irrelevant. BCG Global. https://www.bcg.com/publications/2024/software-projects-dont-have-to-be-late-cos tly-and-irrelevant?utm_source=chatgpt.com