Caveat
Most of this wouldn’t make sense to us either if we weren’t in the eDiscovery industry. If you are friends or family, we help prepare documents for review and production in legal proceedings. If that didn’t help, call us, we always want to hear from you.
Beginnings
Our first client had a simple problem, the forensic collection of a Gmail account didn’t include labels. Without labels, they couldn’t follow their client’s guidance. We identified a key field in the collection that would enable us to use the Google API to download the labels for every email and overlay them into the existing review database. Later, we created a method to collect new emails incrementally, with the client’s permission, without bothering them.
Realizing we could be free from duplicating what so many quality eDiscovery providers already do, we held out for jobs that suited our unique skillset.
Growth
2019: Our first full year managing a law firm’s eDiscovery infrastructure. Our client provided the hardware and software. They identified a gap between what their IT department could deliver and what a software company would support. We took the job of implementing best practices for all and filling in the gaps between hardware, software, and implementation. We focused on:
- redundancy -data can be lost in so many ways, is there always a recovery option?
- performance - what is slowing down the slowest thing, fix it, identify the next slowest thing.
- nagging problems - like our list of missing features, we wanted to identify what was hurting our client, how to get around them, then how to prevent them.
- updates- harddware and software almost always get better. Eliminating bugs and enabling newer features faster maximizes our client’s investment.
2020- EDS.Relapp.DocJobs -Our client created a series of transformations, coalescence, and migration to Relativity objects. Mostly possible by running a series of Relativity scripts but lacking error handling and consistent execution. Could we reduce all the steps to a simple interface and handle the expected errors? Searching across multiple databases, we developed a Relativity App that reduces the process to a few clicks.
2021: EDS.Relapp.Sync -Another Relativity-specific solution to quickly move data from one workspace to another. Our client was moving from an industry-leading processing software to Relativity’s own. Frustrated by the inability to correct most mistakes in processing, they developed a practice of processing data in one workspace and reviewing it in another. Further frustrated by Relativity’s integration Points that took six days and multiple steps to sync data, we created our sync application. We relied on Relativity APIs when performance would allow and injected SQL to SQL when we found bottlenecks. In the end, sync time was reduced to two hours on the test data set and handled multitudes of exceptions. Since then countless terabytes have been synced across workspaces.
2022: Open Source Translation - After a few years of sending document translations to AWS or Azure, a client identified Argos (https://github.com/argosopentech/argos-translate) as an alternative. The opensource project offered neural machine translation while keeping all the data local. We created an interchange that allows translation from 22 languages to english with no per document fees. We added sentence parsing, langugae detection, and a queue to keep Argos running and handling execeptions. Currently, we are working on GPU acceleration of the neural network.
The future
We have an an app in mind, to connect all the data, to all the transformations. Regardless of where your data is, we want to apply all of the modern services to it:
- Classification
- Entity Extraction
- Image identififcation
- Translation
- Sentiment Analysis
- The Next Big Thing
Pick your platform, it shouldn’t have to rely on all of it’s own intelligence. Can we integrate all the eDiscovery platforms with the leading intelligence platforms? That’s what we do when we aren’t helping our clients solve their issues. We look forward to the day when we can connect your data to the best intellgence available.
and more…
Every time an attorney codes a document, they are adding work product. While their work product is specific to the case they are working on, can we use this to start the next case? Neural learning leans heavingly on a training set, law firms are the best at producing them in their normal course of business. We want to start your next case, based on your previous experience. Your work product is paid for and proven, let’s start from the best data available.
