How to Make AI Practical in M&A

Artificial Intelligence has taken the world by storm, and there seems to be no way of stopping it. Every industry in the world has adopted AI, and M&A is no different. The integration of AI is revolutionizing how deals are sourced, evaluated, and executed. In short, AI is becoming an indispensable tool for M&A professionals. In this article, we discuss how to make AI practical in M&A featuring two AI specialists: Michael Bachman, Head of Research, Architecture, and AI Strategy at Boomi, Chris Cappetta, Principal Solutions Architect, at Boomi.  

You still generally want a human in the loop, but if a Large Language Model (LLM) can make a human's process 80 percent faster and better, that's a huge win. - Chris Cappetta

According to these experts, making AI practical means understanding the goals and orchestrating how to make LLMs work for customers in various ways, using RAG, fine-tuning, or agent building. 

Large language models

What makes LLMs useful today is their ability to deal with unstructured data, which represents 80 to 90 percent of all data created. Even with structured data, traditional machine learning requires humans to think like a logical computation system before LLMs. Now, it can emulate human-like thinking. 

Generative AI like ChatGPT can handle many tasks without specific training. They have so much knowledge that they can classify sentiment and more just by following instructions. These AI models not only can filter data, but also create new data. And they become even more powerful with the right logic, sequencing, steps, and extra data. 

Retrieval augmented generation

As good as LLMs, it doesn’t have all the data, and Retrieval Augmented Generation is a technique to improve AI responses. The idea is to help the AI by giving it the information it doesn't have or wasn't trained on. By including the right information in the prompt, the AI can better understand the user's question and respond more accurately.

Fine-tuning

Fine-tuning is like creating personalized versions of an LLM. It can make a model better at specific tasks like classifying or routing information. Fine-tuning is especially good for changing the tone of the model's responses, but it doesn't change the facts the model knows.

Practical uses of AI in M&A

The most practical use case of AI in M&A is the summarization of long documents such as contracts. This will make them them easier to understand quickly. Just feed the documents to the LLM and it will do the task for you. 

However, accuracy is always a concern. The output still needs to be verified by a human. Another is mapping, understanding how one document or process could map to another, even in a deal room.

AI can also be used creatively for tasks like classification, keyword generation, and even voice interactions where you can talk to AI to brainstorm ideas. Some people might even use AI as companions, having conversations with them.

‍