Mergers and acquisitions (M&A) take a lot of work and attention to detail. Datasite helps guide its customers through the complex processes, including steps to redact highly sensitive data before a deal is final. The company wanted to automate and streamline the redaction process using AI while still safeguarding its customers’ information, so it turned to Microsoft Azure Cognitive Service for Language and Translator to enhance its Datasite Diligence application. Using Language Service named entity recognition, Datasite added a new feature, Redaction AI. Now, the lawyers and investment bankers who coordinate sales can reduce redaction times by up to 80 percent, helping to move deals forward faster and support successful outcomes.
“Sellers might have up to 100,000 documents associated with a deal, so the time savings can be absolutely massive. Datasite’s powerful technology platform, now with AI features enabled by Azure Cognitive Service for Language, can potentially compress weeks of work into days for our customers.”
Thomas Fredell, Chief Product Officer, Datasite
Trustworthy due diligence
Selling a company is usually a once-in-a-lifetime event. The company being acquired must execute numerous unfamiliar steps. Datasite offers end-to-end support in this complex process to help all parties in M&A have a smooth experience and successful outcomes. Datasite provides leading web-based software tools for the M&A industry and serves customers in 170 countries.
At the beginning of the M&A process, a selling company will hire an investment banker to coordinate a strategic sale. The banker helps the company curate the sensitive data and documents that are needed for the sale, working with the company’s internal leaders across HR, finance, and intellectual property protection. Eventually, the number of prospective buying companies gets whittled down to a few finalists that need to see this detailed information about the seller to make a buying decision. This is called the due diligence phase of a deal, where the seller must provide essential financial and operational data. This data often includes sensitive business information and personally identifiable information (PII), so many sellers share it in a highly secure virtual data room.
However, because multiple potential buyers have access to the data room, the seller needs to redact the PII for legal and security reasons. The selling team determines the named entities associated with the PII that it needs to redact. Historically, the bankers, lawyers, and other selling team members would then manually comb through thousands of documents to find and address that information. It had to review documents, black out data, possibly print and rescan the documents, and then upload them to the data room.
“Manual redaction was a big pain point for all parties, especially the investment bankers who prepare the deal,” says Ashish Pagey, Vice President of Artificial Intelligence at Datasite. “It’s a very slow, cumbersome process.”
Speeding up redaction with AI
Datasite wanted to automate redaction in its industry-leading Datasite Diligence application to take place quickly, accurately, and securely. The company knew that the process could be streamlined by adding AI, such as natural language processing capabilities. After evaluating various options, Datasite chose Microsoft Azure Cognitive Services because it offered the sophisticated AI and powerful modeling the developers wanted.
The company developed a feature called Redaction AI to augment its Datasite Diligence application. The feature uses Azure Cognitive Service for Language to automatically evaluate the text in all the seller’s documents. Redaction AI draws on the named entity recognition cognitive skill to extract named entities from text using sophisticated AI. Now everyone on the selling side can use Redaction AI to redact named entities in batches instead of going through one by one.
“It’s incredible to think about what redaction was like for customers before and after,” says Thomas Fredell, Chief Product Officer at Datasite. “By having access to these text models in Language Service, we give them the option to batch work that they would have otherwise done individually. Sellers might have up to 100,000 documents associated with a deal, so the time savings can be absolutely massive. Datasite’s powerful technology platform, now with AI features enabled by Azure Cognitive Service for Language, can potentially compress weeks of work into days for our customers.”
Adds Chris Miley, Senior Product Manager for AI ML at Datasite, ”We’ve had customers tell us that Redaction AI saves them up to 80 percent of the time and resources previously needed to complete the redaction requirements of a deal.”
Moving faster increases the likelihood of a more successful sale for sellers because any slowdown to the deal process can damage a potential acquisition. “We have a saying here at Datasite—‘Time kills all deals,’” says Dayo Famakinwa, Vice President of Product Management at Datasite. “If you’re trying to sell or buy a company, external factors such as the economy or a global pandemic are more likely to change the dynamics of the deal the longer things take. Time is of the essence to redact content quickly and to do everything you can to close as soon as possible.”
Serving a wider customer base
Datasite also adopted Translator, part of Azure Cognitive Services, to translate the documents inside a virtual data room. A deal might take place across international borders or have documents in multiple languages, and the investment bankers and others involved might not speak those languages. By having nearly instant access to translated documents in the data room, sellers save both time and resources because they don’t have to have those documents translated.
“The selling team still needs to arrange documents in the right location in the virtual data room even if they don’t speak Spanish or Italian or whatever the target language might be,” says Famakinwa. “Those documents need to be translated—not necessarily to a legally binding extent, but enough so that the team can understand the context and put those documents in the right location. Thanks to Translator, investment bankers who use Datasite Diligence finish organizing and setting up projects very quickly, with no need to engage a third party to understand the cross-border content.”
Adds Pagey, “If we’d tried to support Redaction AI with a custom-trained model, that would’ve taken us a lot more time. Using Azure Cognitive Services with a comprehensive set of language and entity support gave us a head start, and we put our solution into market much faster.”
Accurate, high-quality service to customers
By creating Redaction AI, Datasite sped up the redaction process and enhanced service for its customers across regions while also delivering an extremely high level of accuracy. “We have a set of documents that we use for testing and establishing our interpretation of accuracy, and we’ve seen significant improvement,” says Pagey. “We work on many kinds of projects with a lot of document variation—in multiple languages formatted in a variety of ways. In addition to accuracy, the models in Language Service do a great job handling that variety of data.”
Overall, Datasite is achieving its goal to provide a streamlined, industry-leading service that takes the weight of the M&A process off of its customers’ shoulders. Concludes Pagey, “From the beginning when we were evaluating various options, we found that we gained the most coverage for named entities, languages, and outstanding accuracy with Language Service and named entity recognition.”
Find out more about Datasite on Twitter, YouTube, and LinkedIn.
“If we’d tried to support Redaction AI with a custom-trained model, that would’ve taken us a lot more time. Using Azure Cognitive Services with a comprehensive set of language and entity support gave us a head start, and we put our solution into market much faster.”
Ashish Pagey, Vice President of Artificial Intelligence, Datasite
Follow Microsoft