Could you give any insight into what problems you're trying to solve with the engineering side of Atrium? When you commonly hear about legal tech, its in the context of e-discovery/case-tracking/forensic tools.
When we started Atrium, we thought the main tools would be web software applied to the legal business: CRM, project management, workflow tools, document generation.
What we've learned is that there is a more interesting/difficult technical challenge to what we want to build. I think of Atrium's technology as a platform comprised of several layers: a document storage system (stores legal docs -- think PDFs and Word documents), an ML layer that parses those legal docs and turns them into structured data, an API / web platform layer that exposes that data to applications, and then a bundle of applications that sit on top of that data.
That may be a little complex, so I'll give you one example of this in action, which we currently use in production:
When a startup does a Series A financing, the first step is a VC hands the founder a term sheet. The founder then gives this term sheet to their attorney, who will take it and read all the company's previous SAFEs and convertible notes line by line. They then will create a pro forma cap table (basically, an excel model that shows who gets what at the end of the financing). This process can take a lot of human effort (i.e. the lawyer will spend hours reading the notes and creating the model), and it can often be error prone (because someone is doing it manually).
Of course, any programmer who looks at this process is going to realize it can largely be automated. We ingest those documents into our platform before the founder gets the term sheet. Our models then turn those SAFEs and convertible notes into structured data by mapping them to a schema of what we expect to be in those documents (this is verified after the fact by humans). Then another application just renders the pro forma cap table, taking a process that took hours and reducing it to a matter of minutes. This spits out an XLS, so the attorney can take it and use it to advise the founders (or send to the VC's counsel) just like they would a cap table they built themselves (i.e. we aren't prescriptive on how they advise).
My goal is that eventually we understand every kind of corporate legal document, and make that data accessible to a universe of client / lawyer applications that can streamline workflows and the delivery of service.
