[EN] Breaking the Due Diligence Bottleneck with AI: Finding Key Insights in a Sea of Data

While due diligence (DD) has always been the bedrock of M&A transactions, its role has expanded from being a defensive safeguard to serving as a competitive value-driving differentiator within the global private equity landscape.

As important as the thoroughness of DD processes is the speed at which firms can effectively identify and address key risks in a transaction. Within markets such as private equity that are characterized by extreme asymmetry of information, AI-powered due diligence can provide significant value. It can serve as both a vital tool for managing downside risks early on in an investment, as well as an identifier of value-add opportunities that traditional methods may overlook.

Transforming Time-intensive, Manual Processes with AI

Anyone in private equity is familiar with the scenario: once a deal is launched, analysts need to navigate through an extensive dataroom filled with thousands of documents and complex data sets. Deal teams often need to spend hundreds of man-hours sifting through this massive volume of data for potential red flags. However, it is becoming increasingly evident that relying solely on manual diligence is no longer an efficient or competitive method of processing massive amounts of information.

Centroid established an in-house AI/Data Cell in 2024 to tackle this core challenge – to develop a solution that can help quickly and accurately navigate through significant amounts of data, and address some of the largest bottlenecks and inefficiencies in the deal review process. Following extensive development and close collaboration between the front and back office teams, we recently launched an internal AI-driven platform that is being deployed in live transactions and providing real value. More importantly, the platform continues to rapidly improve and grow through live use cases and user feedback. Internally, employees are reporting real benefits and witnessing a fundamental change in the nature of due diligence.

The Challenge: Fighting the Clock and Data Discrepancies

Data encountered in large-scale DD is often a maze of fragmented or inconsistent, while containing noise that makes it difficult for automated review.

• The Physical Limits of Review: Manually reviewing hundreds or thousands of documents can take weeks. During this process, valuable deal team capacity that should be directed towards judgment and strategic analysis can often be consumed by repetitive review tasks.

• Security and Technical Constraints: Given the sensitive nature of investment data, public, consumer-facing AI platforms were not a viable option due to security concerns. In addition, the need to process large and complex files simultaneously posed meaningful technical challenges.

Beyond the sheer volume, PE due diligence involves significant complexities, such as identifying subtle inconsistencies between management representations and underlying documentation, understanding the lagged impact of historical issues on current performance, or detecting nuanced contractual clauses that may be overlooked by standard checklists. The true bottleneck does not lie with simply needing to read faster, but in needing to clearly identify where material risks and discrepancies are most likely to reside.

The AI Solution: AuditX

In response, Centroid’s AI team developed AuditX, a DD support tool based on secure enterprise-grade LLMs. AuditX is designed to enable deal teams to bypass manual search processes and extract relevant information directly by interacting with an AI agent.

Key capabilities include:

• Smart OCR for Data Ingestion: The system can accurately process any document format, including complex charts or multi-language texts, converting them into searchable, analyzable data sets.

• Intelligent Priority Filtering: AuditX scans the entire dataset to identify and surface the most critical documents regarding business, financial, or legal risks, allowing teams to focus their review where it matters most.

• Evidence-Based Analysis: The system does not infer or speculate. All outputs are fully traceable, with citations and direct references to the source material. This ensures that while AI accelerates analysis, final judgment and verification remain firmly with experienced professionals.

• Quality and Security Control: AuditX operates within a secure, independent infrastructure using enterprise-grade LLMs, eliminating the risk of data leakage. A structured prompt library ensures consistency across transactions, while rigorous grounding mechanisms are applied to minimize hallucinations. The entire framework is governed by internal AI principles emphasizing accountability, transparency, and ethical use.

The Results: Compressing Weeks of Work into Hours

The impact has been immediate and measurable. Initial risk review process that used to require weeks of effort from a full deal team can now be completed in a matter of hours – structuring the data and highlighting key areas in advance for more efficient human judgment and review.

This acceleration has also enhanced our collaboration with external advisors. Armed with AI-generated insights, Centroid teams are able to engage advisors much earlier and with a greater depth of insight into the subject matter, meaningfully improving the overall quality of our DD processes. Analysts are able to spend less time on repetitive tasks, freeing up valuable hours for higher-value, multi-dimensional assessment and judgment.

This change is about more than just speed. Platforms like AuditX allow our teams to analyze documents with greater focus and granularity, enabling a comprehensive deep dive from the very start of our review processes. Red flags are identified earlier, allowing for more precise scoping of diligence workstreams and more effective engagement with advisors. From an investor perspective, downside risk is addressed earlier and more methodically.

• Case Study: In a recent manufacturing transaction, AuditX reviewed thousands of pages of quality and safety records and identified a potential recall risk at an early stage. Under a traditional approach, this issue might not have surfaced until later stages of DD, which may have led to unexpected delays in the process. The early identification allowed us to adjust the scope of our diligence form the onset, and structure specific focus points for our advisors.

The Journey Towards an "AI-Native PE"

AuditX represents more than a technical enhancement; it represents a new standard for how private equity firms should engage with data. With large, complex datasets no longer creating friction, proactive and systematic data utilization becomes embedded in daily workflows.

This innovation was made possible by the close collaboration between our AI/Data Cell and our deal professionals. The AI team identified specific operational pain points through interviews, while deal teams provided continuous feedback to refine functionality. This iterative process enables Centroid to develop and deploy new minimum viable products on a monthly basis, rapidly translating technical innovation into practical value.

Centroid will continue to expand its AI initiatives across portfolio management and reporting processes, and refine investment processes to deliver market-leading value. By strengthening our ability to identify risks earlier on and developing a wider range of upside opportunities, we aim to deliver differentiated outcomes for our investors and portfolio companies. This is our vision of AI-Native Private Equity.