Berry Appleman & Leiden is among the world's largest immigration law firms and growing at a fast pace. As the firm grows, the IT organization is aggressively using cutting-edge technologies such as RPA and AI to automate and optimize legal services.
Under the leadership of Vince DiMascio, chief information and technology officer at BAL, the firm has deployed many AI capabilities. They include machine learning for describing data and predicting outcomes, computer vision for classifying and extracting information from documents and natural language processing for interpreting requests for information. In addition, robotic process automation tools from UiPath, in tandem with the programming language Python and ServiceNow Orchestration, are being used to automate rote processes.
Editor's note: This transcript has been edited for clarity.
How are you utilizing RPA and AI at Berry Appleman & Leiden? Let's start with AI.
Vince DiMascio: We've brought AI into our firm this past year -- and even before this year. We have AI in several forms. One is general machine learning for clustering, describing data, predicting outcomes. Another [AI capability] that we commonly use is computer vision for classifying and extracting information from documents -- we're a document-intensive industry -- so we can use systems to manage that workstream for us.
Natural language processing is another AI capability that we bring to bear across our enterprise. On the legal side, we use natural language processing for interpreting requests for information and evidence and so on from governments and for authoring draft materials to respond to such requests. [We also use natural language processing] for classifying content coming in from our clients and from the foreign nationals, so that we can provide faster service to them.
How are you doing this AI work? Are you using in-house talent, partnering with vendors?
DiMascio: We partner with outside vendors: We use a firm called Accelirate. The tools we use [include] UiPath, which is our RPA platform. We also couple that with Python. And on our IT operations side we have ServiceNow Orchestration, which has an RPA-like capability, to bring together processes across our operations -- in HR, IT, for onboarding, for account management and self-service.
The talent we have in each one of those segments is both contracted and our own people. So for RPA, as an example, talent is scarce. It's relatively new, and the demands are increasing and are high right now. So we are using our outside contractors to help upskill our own people. They learn by partnering with the outside firms. Products like UiPath -- and UiPath itself -- provide great training.
What challenges have you faced in implementing these RPA and AI technologies?
DiMascio: The technologies, generally, are easy to stand up to a proof-of-concept stage. It's fun. It's great. But when you try and bring them to production and impose the controls -- we're ISO certified -- that certifications require, it becomes more difficult and challenging to manage the data. You might have a new repository that you need to subject to scrutiny for regulatory and policy reasons.
Operationalizing [AI technologies] into the business also requires training and socializing what's coming and what it means to people's jobs, which is, by and large, a good thing. The rote work is going to be managed by machines and that frees people up to do more with client service and do more of the work they want to be doing.
Has the firm cut jobs as a result of the RPA and AI tools you're using?
DiMascio: We're in a massive growth mode. We have -- I think -- 200 open requests right now for staff. What we're using RPA for is to push that rote work, the routine work down to machines, so we can scale our legal providers to focus on the professional legal services our clients want from us. Likewise, on our operations side, we want our HR people and our IT people and our procurement people and facilities [people] focused on those [services] and not the routine, so that category of work is being delegated to where it should be -- to machines -- through RPA and related tools.