London-based legal AI company TagDox has formally launched after a period of Beta testing. TagDox, the brainchild of Eli Luzac, uses machine learning and natural language processing (NLP) algorithms to find and tag information inside documents.
The venture, which is backed by several investors including the well-known former head of global law firm Linklaters, Tony Angel, is to some extent a hybrid legal tech company, bringing together several different applications such as NLP document review and collaborative sharing of tagged data to allow faster and more efficient reading and analysis of documents.
In terms of the benefits this brings, in the words of the company: ‘Relevant information is automatically highlighted and can be shared with colleagues, clients, counterparties. This reduces the challenges of document overload and work duplication. By making content retrieval and knowledge sharing easier and quicker, it empowers users by delivering greater insight from previous interactions.’
There are pre-sets of provisions for certain types of legal document, along with an additional ability to add bespoke provisions where needed.
At present the Founder, Luzac says that it is being piloted by at least one global law firm, though understandably he would not comment on who these law firms are yet, as well as an asset manager.
Artificial Lawyer asked Luzac about his invention and where it fitted into the emerging legal AI ecosystem.
‘TagDox is in a different category [to other legal AI companies]. It’s not doing large volume due diligence and it’s not seeking to help provide legal advice. It’s a document analysis platform,’ Luzac explained.
Although it is not designed to be a competitor to the volume contract analysis systems such as Kira and RAVN, it could be used to help reduce the workload when it came to reviewing contracts, whether in due diligence or for other reasons.
This is because TagDox separates out the key provisions of the documents and they are stored and can be shared, in effect building a collaborative knowledge base of ‘tagged’ information. When lawyers return to the documents, or send them to another party, there is less need to do a complete review again as the key provisions and clauses of the documents have already been extracted and organised.
That said, Luzac does not see firms and clients immediately running their vast stores of legal documents through TagDox to provide these tags, rather law firms and other businesses that deal frequently in contracts, will make use of the system on a case by case basis.
Fundamentally, TagDox appears to be an AI system focused on preventing lawyers wasting time from having to read the same documents twice. The aim is that once TagDox has filleted a contract of its key provisions those insights are stored and can be shared from then on.
Less re-reading means more time spent on more valuable work. It also means less associate time spent on matters that clients may not be that willing to pay for. And, as Luzac adds: ‘Because we are using machine learning there is increasing accuracy. The more people use TagDox the more accurate it will be.’
And in terms of future growth, that is in the hands of those currently piloting the system. However, Luzac said he is already impressed with the enthusiasm of those who have tested TagDox.
‘We built TagDox as a legal aid, but have been impressed with the innovative uses our testers found. We had VCs using it for termsheets and even university lecturers and students sharing documents. We’re looking forward to seeing TagDox used in new ways we hadn’t expected,’ he adds.
The UK company currently has a staff of five people with several of its developers based in Tel Aviv, Israel.