AI in Your Pocket? The Potential of Edge Computing to Lead Technological Transformation at Law Firms

Law firms tend to be conservative when purchasing technology solutions—and for good reason. Lawyers are entrusted the most sensitive data that a client can possess, making the cost of a network breakdown or data breach exceptionally high. However, this deep concern for security often comes at a meaningful cost. Many law firms rely on trusted legacy systems for critical processes, meaning that it is impossible for them to deploy modern technologies that could automate countless administrative tasks in their technology stack. In particular, law firms that are not willing to host data in the cloud forego the ability to use most mobile applications currently available in the legal tech marketplace.

A distributed computing paradigm known as edge computing has the potential to change this reality for law firms by bringing computation and data storage closer to the location where it is needed. Edge computing involves processing data directly on a device, without needing to transport data to another server environment. This provides real-time data analysis for the user without compromising data security by minimizing the need to store data outside of the device and the risk of it being hijacked by a bad actor in transit. Although edge computing has been around for over 30 years, it is gaining traction now as the Internet of Things (IoT) becomes more prevalent, creating a greater need to process large amounts of data immediately and in a decentralized way.

While very few legal technology providers have successfully implemented a solution that leverages edge computing, this development has the potential to be revolutionary for law firms, especially as lawyers spend more and more time working from mobile devices. In developing mobile apps for the legal market, the trade-off between security and efficiency will no longer exist—rather, law firms will be able to deploy modern software that is both powerful and extremely secure.

The Advantages of Edge Computing for Law Firm Applications

Security

Traditional machine learning approaches often require data to be centralized in a cloud server. For many law firms, this makes deploying certain AI solutions untenable, since they have often made a strategic decision not to put sensitive data in the cloud—and for good reason. As Capital One’s data breach of 100 million accounts in 2019 demonstrated, transitioning to a full cloud computing architecture comes with a myriad of security risks for even the most sophisticated IT organizations. Often, unauthorized access to data stored in the cloud happens as a result of an IT team’s failure to properly secure company data rather than because of user error. If an IT team is not extremely skilled in deploying cloud architecture, then law firm risks compromising its clients’ information—a cardinal offense in the legal world.

In contrast, mobile edge computing brings intelligence closer to the edge, where the data is produced. With these solutions, IT departments are not required to do as much work to secure the data, since it is not stored on a centralized server. The distributed nature of edge computing also reduces the risk that a data breach can affect multiple devices or allow a bad actor to access data from multiple devices.

Privacy

One of the most critical challenges facing law firm IT departments is not technical at all, but human: user adoption. Lawyers are not known for being avid technology adopters, and this tendency is partially the result of an aversion to risk and sensitivity to data privacy. For example, with one of our very own clients, we experienced a user’s disinclination to using their law firm’s document management system because they were reluctant to share their communications in a firmwide database, even with security measures in place. Ultimately, it can be the law firm that is most hurt when its users refuse to adopt such critical technologies. Consequences can include falling out of compliance with client guidelines and being open to legal and financial repercussions.

Edge computing allows users to maintain privacy by processing data on the device itself and not in a centralized location. Apps powered by edge computing can provide personalization by learning from the user’s data without making that raw data accessible outside of the user’s device itself.

Protecting User Privacy with Edge Computing

There are three common approaches for on-device machine learning algorithms that can be implemented to protect user data:

  1. Train on server and distribute the trained model to the edge (teacher-student model). In this scenario, the machine learning model is trained on general data in external servers (teacher network). This data is then compressed to mimic the output of the teacher network, producing the student network—a distilled model that exists entirely on the user’s device. Apple’s Face ID uses this methodology in order to ensure that its facial recognition feature does not compromise the privacy of its users.

    teacher student model

  2. Federated Learning. Unlike the teacher-student model, Federated Learning (FL) provides a decentralized approach to training algorithms. Using FL, the machine learning model is trained on the specific data source, and the output (i.e. a summary of the user’s data) is then moved to an external server for further analysis. In cases where there are large data sets, companies can then improve their algorithms by using the output of the locally trained machine learning model rather than directly accessing their users’ data.

    federated learning model

  3. Differential privacy. In this model, training occurs in external servers by using both the user’s data and random noise. As a metaphor, think about the model as a scientist looking to perform an analysis on survey data about taboo behavior. The scientist receives the data in aggregate but does not hold direct access and cannot tie a single piece of information to an individual participant. In our instance, the algorithm takes the place of the scientist. While it will learn and improve over time with this generalized data set, it will not be able to compromise individual user data—meaning that for companies, they can use this model to collect and share aggregate information about user behavior without compromising the privacy of individual users.

    differential privacy model

If a law firm is able to successfully drive adoption of technologies that can leverage these models to process and analyze user data on the device and gain user trust, it can open the door for the usage of cloud-based technologies later, when users can feel confident that the sensitive data is processed directly on their device, and anything processed off the device is not sensitive.

How ZERØ Uses Edge Computing

More lawyers than ever are using mobile devices for work, with over 70% of lawyers reporting in the most recent ABA Legal Technology Survey that they primarily use their smartphones to access email outside of the office. However, until now, no technology solutions have existed that have allowed lawyers to optimize this email usage from a mobile device. This means that much of the time lawyers spend working from mobile devices goes uncaptured (and therefore not entered), leaving law firms with untold amounts of lost revenue. In addition, in order to properly comply with document retention guidelines, lawyers must either screenshot their emails to file them into their document management system or just wait until they are at their desktop to do so, creating a risk that they might forget.

ZERØ is currently the only legal technology solution on that market that implements edge computing to solve these problems and provide five major benefits to its users:

  1. Predictive and automatic email filing into iManage and NetDocuments, so lawyers can file emails as soon as they receive them, wherever they may be.
  2. Passive, automatic, and contemporaneous time capture of interactions with client-related emails on mobile devices, so lawyers can worry less about manually recording this time later.
  3. On-device narrative generation, so that precise time reports related to interactions with client email on mobile devices can easily be entered into time entry systems.
  4. Real-time potential wrong recipient detection—before the email is sent.
  5. Supercharged productivity filters that allow users to see the information that is most important to them at the top of their inboxes.

Because ZERØ processes user data directly on the user’s device, there is no complex configuration required with your firm’s technology stack, and users do not need to worry about their sensitive data ever leaving the device. ZERØ’s architecture also makes us cloud-agnostic—meaning that our mobile app and Desktop Companion for Outlook will work for your firm whether or not you have moved your applications into the cloud.

Are you interested in learning how ZERØ can help your lawyers be more productive and your law firm be more profitable? Request your demo today.

Tags

Gevorg Karapetyan

As ZERØ’s co-founder and CTO, Gevorg leads our long-term technology vision and is responsible for the direction, coordination, and delivery of technology. Gevorg holds a Ph.D. in Computer Science and has more than 10 years of experience in developing intelligent automation systems.