The software development unit at the digital law firm Synch is branching out and will be run as an independent company. CTO and co-founder Sergii Shcherbak outlines the plans for the new venture.
The new company, called Maigon, offers GDPR-related legal AI tools, solutions for document automation and data labeling as well as legal AI development services and workshops.
What was the reasoning behind founding a separate company?
- We at Synch have always been passionate about innovation and the value that AI can bring to the legal profession. Having developed and launched such tools as PrivacyPolicyCheck.Ai, which conducts AI review of privacy policies, and Maigon DPA (previously known as DPA AI), a corporate solution accelerating review of data processing agreements using neural networks, we have decided that it is more practical and flexible to branch out the software development unit as a separate venture. This is how the concept of Maigon came to life, says Sergii Shcherbak, who's been managing Synch's software development unit for years.
It's fair to say the reception of PrivacyPolicyCheck was pivotal for this development then?
- PrivacyPolicyCheck.Ai is the project that our journey to real practical AI started from. It was our first take at a deep learning product, which we developed entirely in-house. And we think the service succeeded - it’s being used on a daily basis. Today our AI reviewed privacy policies from Brazil, India, and Malaysia, while yesterday submissions were from the UK, Germany and Sweden. Having seen a proof that our AI works and brings value, which is backed by positive user feedback, we developed Maigon DPA, which is now used by large corporates like Dustin and already saved hundreds of hours of legal review time. The positive perception of our AI solutions by businesses and the rising interest in these and other AI tools has majorly contributed to the decision of founding Maigon.
And apart from providing tools you will also act as consultants or developers for other organizations?
- Absolutely. Maigon is an independent legal AI company, and we are ready to collaborate on exciting LegalTech projects with businesses. We are always looking for partners with whom we can bring new bright ideas to life. For us it is important to contribute to the development of Legal AI globally, and most exciting things happen in collaboration.
What markets are you targeting?
- We think that even though AI is evolving in some regions faster than in others, this technology will ultimately revolutionize the legal profession across the globe. Since the launch of PrivacyPolicyCheck.Ai in mid-2018, more than 3700 privacy statements were submitted for AI review, from 72 countries. This already shows a clear interest and demand in legal AI tools in almost a half of the world’s market. For corporate solutions like Maigon DPA, our first clients are big companies based in the EU, but we see the potential to go beyond that, since the impact of the GDPR is global.
- In LegalTech, in particular in Legal AI projects, data is key. And correctly and carefully collected and labeled data is gold. It is even better when these processes are done by people who have AI training experience and know how AI will “react” to all sorts of data. Therefore, if there is an exciting project on the horizon, we are happy to help with consulting on data collection and labeling as well as doing the “heavy lifting” ourselves.
The Covid-19 tool - how does it work and who is it for?
- During the past weeks, for obvious reasons Synch faced a rapid increase in questions regarding force majeure provisions in contracts. The Covid-19 contract tool, which can be found on the Maigon website, is free to use and helps to quickly identify force majeure clauses in any types of contracts. Businesses can submit, say, a 50-page agreement and find the relevant wording in a matter of seconds, including in cases when force majeure is not directly defined in contracts as such but is worded a bit differently, for example as an unforeseen event or similar. I think it is quite helpful, when time is of the essence and one needs to check a few documents at once. In addition, this tool demonstrates a small part of, and sort of a glimpse into, our technology.
Do you develop everything in-house, or do you integrate external technologies provided by other AI vendors in your products?
- It is a good question, since indeed some legal AI solutions out there are more of an abstract-level and based on the API provided by other vendors. And it is not always a bad thing, since the most important is the value the product brings to the customer.
- As for our solutions, the AI has been developed and trained by us, using internally collected and labeled data, and is fully proprietary. We do not make use of services of other legal AI vendors. Back when we started developing our first AI solutions, we quickly realized that it is faster and more efficient to implement all the necessary tools ourselves, and these usually fit our business case better and are more customizable since we have full control over the code. Due to this customizability and control, our solutions end up to be more accurate, and therefore bring more value to the table.
- But of course, we “stand on the shoulders of giants” when it comes to the base-defining open-source libraries and architectures for NLP and machine learning, like BERT, TensorFlow, Keras, PyTorch, Transformers, spaCy, NLTK, scikit-learn and others. Another important aspect is the infrastructure, which must be scalable, reliable, and be able to handle the increasing workloads. We use Amazon Web Services (AWS) as the infrastructure of choice.