Prepare Drafts With Precision and Confidence Using AI-Powered Draft Analyzer
Bloomberg Law’s new Draft Analyzer tool was conceptualized and created by lawyers, for lawyers. Senior Product Manager Andrew Gilman and Legal Analyst Grace Maral Burnett, both of whom have backgrounds as M&A attorneys, share details about how the tool works and how it can help lawyers at all levels work through the drafting process with more efficiency and confidence.
[Learn more about how Draft Analyzer uses AI and machine learning to benchmark draft language against the market standard.]
What is your background in the legal industry?
Andrew Gilman, senior product manager: I started as an M&A attorney about 21 years ago. As a junior associate, I was doing a lot of due diligence and document drafting. So I was pretty familiar with many of the common annoyances in the M&A drafting process. When I came to Bloomberg Law in 2012, I thought, “These are issues that we’re capable of fixing with the tools that we have here.”
Now I’m a product manager. The parts of the site that I generally have responsibility for are related to the workflow of in-house counsel and transactional associates and attorneys. I work on a lot of the SEC-related information that we have on the site, EDGAR, and the tools and functionality that we have for in-house counsel and their workflow.
Grace Maral Burnett, legal analyst: Before joining Bloomberg Law two years ago, I spent about four years as an M&A lawyer in Istanbul, Turkey, working primarily on cross-border and private equity deals involving North American and European companies investing in Turkey. I worked on all stages of M&A deals, from drafting NDAs to conducting closings. As far as the contract drafting and negotiation part of the M&A deal process, given the challenges that the Draft Analyzer tool addresses I, as an associate new to M&A, would have very much loved to have the tool. That’s an understatement.
I’m now a legal analyst at Bloomberg Law on our Transactions team. I analyze M&A market and legal trends, and I also write Practical Guidance for M&A and transactional practitioners.
[Perfect your drafts by comparing them to similar documents with Bloomberg Law’s AI-driven Draft Analyzer tool. Request Demo.]
Who is the target user of the Draft Analyzer tool?
Burnett: I think anyone working in M&A, with any amount of experience, can make use of this tool. It’s valuable for associates, but it’s also valuable for more senior-level lawyers who want to have quick insights into the market. You can do what you need to do faster, and you can triple-check yourself, essentially.
Gilman: Right, in short, it’s for any lawyer who is involved in the drafting, negotiating, reviewing, or reading of an M&A agreement. When we designed the tool, the user that we originally had in mind was an M&A associate, the person who gets tasked with putting together the initial draft or renegotiating the agreement – the wording in the agreement, redlining it, sending it to the other side, receiving their comments back, and so on. When we designed the tool, it was that workflow and that particular set of user problems that we were focused on.
That said, there are many features on this tool that we realized would be useful for pretty much anyone of any level who’s working on an M&A deal.
How would you describe the basics of the tool?
Gilman: When a transactional attorney is drafting a document, they don’t start from scratch. No one starts from a blank page; they take a similar document that was used in another deal and modify it to address the specific terms of the deal that they’re currently working on. So there are textual similarities in a lot of these documents that get filed in the SEC’s EDGAR database, but they’re not exactly the same.
When you upload your document to our system, we figure out where your clauses are, and then we classify each clause. We have about 200 different clause types. What that allows us to do is provide granular Practical Guidance of the sort that Grace writes, about the things that you need to consider for a particular clause. What is the purpose of the clause? What are the perspectives of the buyer? What are the perspectives of the seller? We bring that information and these great Practical Guidance materials directly to the user’s workflow.
It also lets you compare your text to the market. We take each individual paragraph and find everything else out there that’s close to yours, and we do a one-to-many comparison of yours against everything else we have. You can see how your language stacks up against all those other documents that are similar. Maybe you’ll find most of these other documents add a materiality qualifier, and yours didn’t. Maybe that’s something you’ll want to consider doing.
Additionally, we calculate what the market standard is for every clause type. We accomplish that by extracting that same clause type out of all these other documents, cluster them together, and identify the most standard version of that clause. And you can filter that down as well. You might not care about the most common version of the clause overall; you might only care about it in deals in the pharmaceutical industry that are valued at $500 million or more. We allow you to filter your results, so you can benchmark your language against agreements from those deals that are most similar to your own.
What type of technology goes into making Draft Analyzer run?
Gilman: We have a bunch of different machine learning models that we use to process an uploaded agreement. For example, we have a machine learning model that figures out where the clauses are and what the clause outline looks like. For every clause we figure out: Is it a clause? Where does it start? Where does it stop? And where does it live in the hierarchy? Is it a subsection of this other clause, or is it a sub-subsection of that clause? And so, we are able to essentially build a clause tree, which is an interactive table of contents, and which powers your ability to navigate the document.
On the left side, we have that tree, where you can click the individual clause and it navigates you to that particular section. We also take each of those clauses and we run them through a different machine learning model to classify them by topic. And we have a heuristics model, which is a rule-based model that pulls out the defined terms and matches them to one another.
One of the most interesting and unique aspects of the new M&A enhancements is the semantic analysis engine that we built from scratch. It’s what allows Draft Analyzer to compare your language to tens of thousands of publicly filed M&A agreements and identify any text with a similar meaning. This allows us to both offer really useful apples-to-apples comparisons at the paragraph level, and to identify market standards at the clause level.
[Download our report for in-depth analysis of the hot topics that are shaping the future of the legal industry.]
How would you have found Draft Analyzer helpful as a new lawyer, and how will it help your work today?
Burnett: As a young lawyer new to M&A, it would have helped me gain a better understanding of drafting nuances, the pro-party considerations in complex provisions, and market standards specific to the particular type of deal I was working on in a much more efficient way than I could have done on my own. In my current work on both analysis and Practical Guidance in the M&A space, I keep close track of developments in drafting trends, so the ability to compare such large numbers of documents helps me do that.
The market standard feature is a big deal. M&A lawyers spend quite a bit of time analyzing and rationalizing negotiated provisions. When they go to the negotiating table, they want to have a rationale for the provisions in their contracts. And sometimes, they even need to be able to explain what they’re going to try to get their client to accept. So they try to do, in a manual way, what this tool does for them. They’ll look at as many precedents as they can to get a sense of what the market standard is and how they might be deviating from it. This tool is taking thousands of agreements and doing that same analysis to a degree that a person on their own just could not do.
It saves you time and it offers you more precedents to compare to. When you are going to the negotiating table, you have more confidence and an almost scientific rationale you can bring to the table. You can say, for example, “We looked at other pharma deals valued at between $500 million and $1 billion, and we looked at this clause type, and this language that we’re proposing is also found in 2,000 other agreements.”
It helps you do what you need to do in a better and faster way, but it also helps you in the negotiating process and adds confidence to the drafting decisions you make.
Reference Shelf
REPORT: M&A Due Diligence Issues Reporting Template
Three Ways to Draft M&A Agreements with Confidence
Artificial Intelligence for Lawyers Explained
Bloomberg Law 2023: Future of the Legal Industry
Bloomberg Law 2023: Transactional
2022 Outlook on Privacy & Data Security
The Asset Deal Trends to Watch in 2021
U.S., U.K., EU Want Sharper Eyes on Big Tech Mergers