First Newsletter - Lumios Law, AI-first eDiscovery!

Hey everyone - since this is our first-ever newsletter, we'll start off with a bit of a refresher on who we are. We're Dhruv and Arnav, AI researchers at Berkeley, now building Lumios Law, an AI-first eDiscovery platform to help litigators and in-house counsel accelerate their discovery processes.

We may have spoken to some of you a few months ago when this was a research project just starting out, or you may be a supporter we've worked with for a while. We're aiming to run a newsletter every month to keep folks posted on what we're up to since many people asked us to keep them updated about what we work on! We'll keep them short, interesting, and include all the major updates from the last month.

Some catch-up

• We've turned this into a company – officially Lumios Holdings, Inc.
• We've raised some funding from some incredible investors (SV Angel, Josh Browder, and many others supporting us in bringing our vision to the world)
• We've started helping a couple companies and boutiques with live litigations!
• We've built out a full product! We can already connect to major eDiscovery vendors, process 1M+ files, build chronologies, build an order of proof, and find critical facts and documents in seconds instead of hours.

Now, into the major updates for the month! This month, we built out some incredible updates to our product, and we'd love to hear your thoughts!

For Litigators

You can now upload 1M+ documents, have them all finish processing in <30 mins, and start building real work product in minutes! Start by uploading pleadings, discovery materials, and other collateral documents, and then the system will build an understanding of the case.

Beyond chronologies, you can build comprehensive profiles for every person and entity in your case:

Next, the AI also pulls all of the claims and parties out of the initial pleadings and discovery materials, and is able to put together a skeleton order of proof with suggestions on facts that support and deny elements of claims. These can be edited and added to as needed as the case develops over time. Note that claims are enriched with elements from our statutes database and linked for reference.

Visualize the relationships in your case with our interactive knowledge graph:

Lastly, we've heard that finding facts and documents in massive cases is a real pain. So we made a case assistant that understands your case:

For In-House Teams

Since a lot of our technology needed to review large numbers of documents at scale is also very useful with responding to subpoenas and requests for production, we built a subpoena suite. Start by uploading a subpoena or request, generate objections, find all the relevant documents, and tag them. Export privilege logs, responses, and keep materials defensible to the court by using our tools for recall sampling, accelerated QC/validation with keyword/regex searches and entity recognition, and documentation of each step (LLM reasoning trail, prompts, human reviewer decisions, statistical sampling results, etc.).

Lastly, for a service offering we are piloting alongside an independent senior litigator: in certain matters where teams find additional litigation expertise helpful, we can facilitate access to a highly experienced litigator who works closely with our platform, reviews its outputs, provides strategic oversight informed by 20+ years of litigation experience, and helps clients oversee outside counsel as appropriate. This model allows our platform to deliver efficiency and scale, while giving in-house teams the option of senior litigation judgment grounded in understanding of how the product is actually used in practice.

Best,
Dhruv & Arnav