The sign-to-close phase in M&A can be difficult, and often fails to meet expectations. Tight timelines, limited data, and risk at every turn are common challenges. This is where AI & machine learning can make a real difference, especially when it's guided by people with business expertise (read more in our first article here). AI-enabled due diligence can bring accelerated clarity on the first 100 days, and a better path to value creation.
The first 100 days include multiple competing priorities in the areas of leadership and strategy alignment, organizational structure and people management, cultural integration, operational alignment, commercial integration, and the integration of back-office functions. Deal teams leveraging the assistance of AI can more quickly identify and prioritize value creation opportunities across those activities.
One big win involves the use of AI to identify and optimize commercial integration and finance functions simultaneously by discovering hidden opportunities that will generate cash and reduce working capital. As we pointed out in Harvard Business Review ‘delayed and ineffective commercial integration can turn a good deal into a loser, because sales growth ultimately determines whether a merger achieves its value-creation goals.’ Typically, a deal team will identify the most obvious, most achievable value creation integration opportunities, but the process is often constrained by such factors as short timelines and lack of quick data discoverability.
Cash Optimization and the 100-Day Plan
The goals of cash optimization are the acceleration of cash receipts, ensuring that cash disbursem*nts are aligned with contract terms and the development of a highly accurate cash forecast. These are critical activities during the first 100 days. Done right, cash optimization will ensure liquidity and will accurately identify the finance levers available to improve cash generation.
Unfortunately, in the construction of a 100-day plan, it’s not always realistic to troll through hundreds or thousands of supplier contracts in just a few weeks, because the endeavor is labor-intensive and time-consuming. Fortunately, with a carefully crafted approach augmented by traditional technology and generative AI deployed in the due diligence phase, the task becomes much more achievable. Large language models (LLMs) by design work exceptionally well with text content, meaning the extraction of complex language from documents and reasoning about that text comes as an out-of-the-box feature.
This is a simple but powerful concept. Deal teams simply specify what they want to gather from a contract—for example, payment terms—and run the expression across all contracts in minutes. Because of the dynamic nature of LLMs, nothing needs to be hard coded; the descriptions are in natural language. Such a solution can extract key terms from supplier and customer contracts—all of them, not just the biggest—and give deal teams insights they can use to optimize cash outflow, accelerate and improve cash inflow, enhance cash flow forecasting, and reduce idle cash.
LLMs have to be managed by a team with skill and experience. Not only do the models need to be given the most effective inputs, or 'prompts,' they need to be surrounded by the right people, process, and technology components in order to be successful. This additional infrastructure is important for key features like providing citations, to minimize the occurrence of so-called 'hallucinations', or nonsensical answers. These systems will continue to evolve as the business, technology, and contracts do.
Although contract analysis is useful across many areas, it has specific applications in cash optimization as follows:
- Accounts payable (AP): The use of an LLM to extract key fields and contract clauses can accelerate the identification of undesirable payment structures of suppliers based on contract terms. An LLM can also pinpoint opportunities to renegotiate supplier contracts by comparing terms in similar contracts, thereby providing a thorough and faster alternative to manual review.
- Accounts receivable (AR): With AR data, an LLM can identify where customer terms have not been enforced, such as payment terms, discount terms, and fees. An LLM can also determine which customers to contact to accelerate payment and reduce days sales outstanding (DSO). With a rich and organized data environment, the LLM can be fed historical data to identify habitually late customers, suggest effective communication channels, and even produce scripts based on prior customer interactions. That kind of outreach campaign, informed by a customer lifetime value analysis can unlock significant value creation opportunities.
- Treasury: Cash forecasting can be improved by extracting supplier payment structures and customer payment behaviors, creating richer datasets for analysts toreview. Eventually, an LLM will be able to assist analysts in identifying anomalies or opportunities.
The Essential Role of Expertise
An LLM is a powerful tool—but still a tool. We have had success in augmenting the ways due diligence and integration teams work not by offloading entire workstreams to AI but by using LLMs where they excel in accelerating parts of a team’s own analysis process. Teams get better data more quickly, which enables them to focus on higher-value tasks and generation of insights.
Cash optimization requires a combination of highly structured data, like an accounts receivable database, and unstructured data like customer and vendor contracts. Before LLMs there were only hard-coded and cumbersome solutions for extraction of useful information from the latter, or they were ignored entirely. Now, we have the capability to extract value literally locked away in obscure PDFs.
AI-Powered First 100 Days Starts in Due Diligence
Delivering an AI-powered first 100 days should be a seamless extension of the AI platform used in the due diligence period. The major components that have to be in place for an AI-powered first 100 days are:
- An AI platform with LLM access: If AI was deployed for due diligence, then extending to this document review use case is relatively simple: a centralized document repository.
- Ideally all contracts should be stored in a central repository that the LLM can access.
- Documents are not restricted by format or language. A properly architected AI platform can process Microsoft Word, PDF, JPEG, and many other formats.
- Likewise, a properly architected AI platform can reason about terms in most languages. An LLM can hunt for a list of terms and clauses it should find and analyze and is developed by business experts working in tandem with technologists.
- The more detailed the given context for a term, the more optimized the written prompt.
- API Integration: An application programming interface (API) that calls to downstream processes and can consume the entire, or any portion of the list of terms.
- AI-powered cash optimization is an end-to-end process solution. To realize the most benefit, the output should be made available to downstream consumers who could make use of it.
- Downstream integration points could include databases, applications, and flat files.
Conclusion
AI-enabled 100-day planning is here to stay: Investors and advisors are increasingly using it to enhance their due diligence, planning, opportunity identification, and value creation efforts with better and faster insights. Looking ahead, we see AI, large language models, and machine learning becoming natural parts of any M&A process.
