Find out what opportunities there are to benefit from the use of generative AI in M&A transactions.

M&A processes are demanding and often time-sensitive. The phases of pre-contract negotiations (Letter of intent), due diligence (financial, legal, tax, commercial, carve-out, etc.) and SPA/APA negotiations, in particular, are characterised by intense rounds of coordination between the seller and potential buyer, and large amounts of information are exchanged. In the age of generative AI, these time-consuming processes should be reconsidered and optimised with the help of AI.

There are opportunities to accelerate these Q&A and negotiation processes (especially in the particularly time-sensitive due diligence phase) using AI-based solutions.

Imagine that at the end of a (sometimes hours-long) physical or virtual due diligence or negotiation meeting, you automatically have a detailed transcript of the meeting, including a summary of the key points, and the action points are already fully listed out (broken down by addressee, if applicable). In addition, the participants can be shown relevant information on the previous meeting content during the meeting (in real time as a function of the AI system).

The benefits of a generative AI solution that can do this are enormous. The time required to take notes and create complete action points (such as information request lists etc.) is considerably reduced, which speeds up the coordination process significantly. However, the benefits go far beyond producing the usual summary of meetings.

How does this generative AI solution work?

In simple terms, AI based note taking is made possible by AI-based speech-to-text models, which are combined with an advanced Natural language processing (NLP) model in the use case described here.

Due to the abstraction capability of machine learning models, it is possible to interact with the stored conversation content during an AI-accompanied meeting and ask precise questions about the meeting content (examples include specific context-based questions on which position a certain person took or which suggestion they made or to list in a table the advantages and disadvantages mentioned in the meeting, etc.; then automatically giving verbal answers to existing written Q&A lists would also be possible).

Even sentiment analyses, as they are known, (i.e. assessments based on quantitative machine learning methods, such as SVM, KNN, logistic regression or RNNs/CNNs, as to whether the statements in the meeting were predominantly positive or negative) are possible through the use of machine learning algorithms.

The main benefit of such a generative AI solution is that it accelerates coordination processes (especially the exchange of Q&A) and reduces possible coordination difficulties and misunderstandings in advance (e.g. in cross-border deals due to any language barriers) due to the fairly high degree of accuracy of such AI-based information processing.

Respecting data privacy

However, the use of this kind of generative AI solution may also require a legal review (e.g. consent of all those involved in the meeting) and could also be ruled out by participants’ data protection and security concerns.

In practice, however, data protection concerns can be addressed by pausing or stopping the AI model’s data collection on particularly sensitive discussion content, for example, or by only starting it at the end of the meeting, at which point all the parties then verbally state the most important points and the action points so that the generative AI can only use this to provide  its added value.

However, the completeness and accuracy of the AI-generated content and the results must still be checked by the user. In addition, NLP models today still have weaknesses in interpreting linguistic peculiarities such as irony and sarcasm and can make mistakes just like humans. However, due to the high degree of automation, it can be assumed that such AI systems are in many aspects comparable to a manually written memo and are clearly superior in terms of speed and the functionality described.

AI can reduce language barriers

In cross-border and inbound transactions in particular, AI-supported messaging can also reduce the misunderstandings due to language barriers. Worldwide, cross-border and inbound corporate transactions are predominantly conducted in English, as in some cases purely national processes are due to the involvement of international private equity investors. In this respect, current deep learning models (neural networks) achieve an unsurpassed recognition rate, even for different accents. As a result, the advantages clearly outweigh the disadvantages, especially in this particular case.

The application of this kind of an AI-based approach is also transferable to any form of business meeting. But the more time-sensitive implementing the underlying business issues is (as in the specific example of M&A transaction processes), the greater the benefits.

Furthermore, the application of these AI-based options is not just limited to virtual meetings but can also be applied through edge computing or edge AI to face-to-face meetings. In the case of edge AI, AI algorithms and models are executed on local devices to enable real-time AI-generated content close to the data source, in this case as the AI assistant of a physical business meeting. In addition, edge computing/AI even offers the advantage of higher data protection and data security, as the data are only collected and processed locally and, if desired, not transferred to a cloud.

Do you want to know how you can benefit from these technological innovations on your transaction projects? Feel free to reach out to me!

Grant Thornton’s Transaction Advisory Services team is at your side to support you in implementing your transaction plans and objectives in these challenging times and to achieve your goals together.