The other half of ArsLex.
ArsLex Associate turns Microsoft Word into a cite-anchored drafting workspace. The Discovery Suite is its counterpart for the record itself: the place your firm collects, reviews, and produces documents in litigation. Same tenant isolation, same encryption, same custody promise. You can run either on its own, or both together.
Everything a matter needs, in one platform.
A firm-run alternative in the same category litigators already know for review and production, without shipping your evidence to a third-party host.
Ingest & load files
Load native files and existing productions. Import and export are compatible with the DAT and OPT load-file formats used by Relativity and Concordance, so data moves in and back out cleanly. The custodian is recorded at the door, whether documents arrive in a load file or straight from your own collection.
Review & coding
Code documents for responsiveness, privilege, and confidentiality designations. Reviewers work from assigned batches, check them out, and hand them back. Keyboard-driven on the desktop, and touch- and swipe-driven in the browser on a phone or tablet.
Bulk coding, with a preview
Code a whole search result in one action. Before anything is written, a preview shows what will change, which family members come along, which families would end up split, and every document where the AI's read of the record disagrees with the call you are about to apply. The lawyer still decides. If it comes out wrong, one action puts it back.
AI-assisted first pass
An AI pass reads each document against your criteria and proposes a call for a lawyer to confirm or override. On a long document it now reads the passages that bear on your criteria rather than whatever happens to be at the front, and it says on the record which pages it read and how much of the document that covered.
Email threads
Review the message that carries the whole conversation instead of reading the same quoted text nine times. Code the inclusive messages and the rest of the thread follows, with a record of where each decision came from rather than a silent fill-in.
Near-duplicates & concept clusters
See which documents are textual near-duplicates of one another, and group a production by subject so related material is reviewed together instead of scattered across the queue. A quality panel surfaces the documents that fit nowhere, which is often where a bad scan is hiding.
Search, by word and by concept
Keyword and phrase search across the whole set, plus search by concept, which finds documents about a subject even when they never use your words for it.
Batches & reviewer QC
Carve the unreviewed remainder into batches by topic or by size, watch throughput per reviewer, then sample completed work for a second look. The sample can be weighted toward the disagreements that carry the most risk, and every report says in plain words which method drew it, so the check can be reproduced.
Search-term & validation reports
Run search-term reports to size a set, test a negotiated term list, and document hit counts for the meet-and-confer. Statistical validation sampling puts a defensible number on what the review did and did not reach.
Privilege screening, PII & logs
Screen for privilege, generate the privilege log from your coding decisions so the log and the review stay in step, and check the roster of everyone being treated as counsel. A personal identifier scan flags Social Security and account numbers before anything is produced.
Redactions
Redact privileged and sensitive content before it leaves your walls, carried through to the produced set.
Bates numbering & gap checks
Apply consistent Bates numbers and endorsements, keep an email and its attachments inside one continuous range, and run a gap check before you certify a range to opposing counsel.
Productions, served and frozen
Assemble and export with a load file, Bates ranges, and redactions applied, tied back to the requests they answer. Mark a production served and the set freezes: what you handed over stays exactly what you handed over, so months later the record still matches the certificate.
Clawback, in both directions
Send a clawback notice on your own production, and take one in on theirs. Enter the Bates ranges opposing counsel is clawing back and the matching documents drop out of review, search, and export. If a range does not fully match what you hold, the shortfall is reported to you rather than quietly rounded off.
Exhibits, and the protective order
Build exhibit lists from produced documents. When a produced document is cited in a brief drafted in ArsLex Associate, its confidentiality designation travels with it, so a document marked Confidential or Attorneys' Eyes Only announces itself before it reaches a public filing.
Processing exceptions in plain sight
Files that could not be processed are listed where you review, with a count on the tab, not buried in a log nobody opens. Password-protected Office files are detected and reported as exceptions rather than passing quietly as empty documents.
Legal holds & custody
Place matters under a legal hold that suspends deletion, and keep the working copy of your production hosted, live, and searchable for as long as the matter needs it.
The review room, on screen.
Real screens from the Discovery Suite. Hover to enlarge; click any screenshot for a closer look.
From collection to production, in one place.
The Discovery Suite follows the workflow your team already runs, from getting documents in to handing a production out, without the data ever leaving infrastructure you control.
From collection to review
- Ingest native files and load-file productions (DAT / OPT), with metadata and custodian preserved. Anything that would not process is listed as an exception you can see.
- Organize by custodian, by email thread, by family, and by subject, with EDRM near-duplicate flagging and coding-consistency checks.
- Review and code for responsiveness, privilege, and confidentiality, from assigned batches, at the desk or on a phone.
- Code in bulk when a whole search result gets one call, after a preview that shows you what changes and where the AI reads it differently.
- Search the whole set by word and by concept, and run search-term reports to test and document a term list.
- Check the work: sample completed batches for a second look, weighted toward the disagreements that carry real risk.
From review to production
- Screen for privilege and personal identifiers, and redact what must not leave your walls.
- Bates-stamp the set, keep families in one continuous range, run a gap check, and build a privilege log straight from your coding.
- Produce with a load file, Bates ranges, and redactions applied, ready for opposing counsel.
- Serve and freeze: mark the production served and the set locks, so what you handed over stays what you handed over.
- Claw back on your own production, or take in an inbound notice by Bates range and pull the matching documents straight out of review and export.
- Hold and host: place a legal hold when you must preserve, and keep the production in a live, searchable warm archive when the matter goes quiet.
Your evidence never leaves your custody.
Three things, and they are worth stating separately. Zero-retention AI: where AI assists, text generation runs on Amazon Bedrock under AWS's zero-data-retention and zero-operator-access terms, never used to train models and never retained by the AI provider, and the search embeddings behind concept search are computed on infrastructure we control and never leave it. Encrypted custody: we hold the working copy of your production, and we hold it under encryption. Provable deletion: when it goes, it goes, and you can see how.
- Isolated per tenant. Each firm's data is walled off from every other tenant, enforced and tested continuously.
- Encrypted in transit and at rest, with access behind authenticated login, and optional multi-factor authentication your firm can require for everyone.
- An encryption key of your firm's own. Documents in our durable store can be sealed under a key that belongs to your firm alone, held in AWS Key Management Service and used for nothing else. Ask us to turn it on for your account.
- Cryptographic deletion, and never a quiet one. Because the documents are sealed under that key, destroying the key makes them unreadable wherever they are stored. Nothing about that happens silently: every administrator is warned in the app and by email, someone has to confirm it in so many words, and a waiting period runs before anything is destroyed. If we cannot reach an administrator, the request cannot go forward at all.
- Backed up nightly, because as custodian of the working copy of your production, durability is our responsibility.
- Long jobs have to prove they are working. A stamping, export, or indexing run that stops making progress is failed and reported rather than left spinning behind a status bar that says it is fine.
- Kept until you delete it. Nothing is on a deletion timer unless you set one, and legal holds suspend deletion for a matter entirely.
- Warm archive, not cold storage. Closed matters stay live and searchable, so a production is never more than a login away.
Usage-based, priced per matter.
Discovery pricing scales with the volume you host and process, and is quoted per matter so it fits the case in front of you. Tell us about your matter and we will put a number to it.
The per-seat plans on our pricing page are for ArsLex Associate, our drafting product. Discovery is a separate product, priced by usage.
Ready to run your next matter on ArsLex?
Sign in to the Discovery Suite to get started, or talk to us about migrating an existing matter.