Discovery questions have a shape: what did this custodian send that person about this topic, in this period. Search on the platform is built around that shape - not around a generic search box that treats a litigation corpus like a website.
Two Engines, One Ranked List
Every query runs both semantic (meaning-based) and keyword search. The two result sets are fused by reciprocal-rank fusion and then reranked by a precision model, so the top of the list is worth an attorney's time. If any stage is ever unavailable, search degrades gracefully instead of failing - a query always returns results.
Filters That Match How Attorneys Ask
- Custodian filters: who sent it, who received it.
- Date ranges, mailbox labels, and source type - email, document, image, transcript.
- Every AI-coded facet: issue tags, helps/hurts bearing, hot documents, privilege flags, and references to specific paragraphs of the operative pleading.
- Inline operators typed straight into the query -
from:,to:,after:,before:,kind:,label:,is:hot,is:privileged- with the interface confirming which operators it applied.
From Hit to Context in One Click
Results are grouped per document with the best passage highlighted and further matching passages one click away. From any result you can open the full document, jump into its email thread, or pivot to "find similar" documents - so one good hit becomes a trail, not a dead end.
Noise Control Without Data Loss
Newsletters and bulk mail are screened out of ranking by default so they stop competing with evidence. But the screen is a ranking decision, not a deletion: everything remains in the index and is searchable on demand. Nothing is thrown away to make search look cleaner.
Works Together With
- AI Issue CodingEvery document coded for issues, helpful or harmful bearing, privilege, key quotes, entities, amounts, and dated facts.
- Document ProcessingEmail archives, documents, spreadsheets, scanned paper, images, audio, video, and nested archives - all made searchable in one index.
- Case IntelligenceA cited case chat, chronology, hot and adverse document queues, truth table, entity graph, and per-claim strategy pages.
Frequently Asked Questions
What is hybrid search, and why does it matter for review?
Every query runs two engines at once: semantic search, which matches meaning, and keyword search, which matches exact terms. The results are fused into one ranked list and then reranked by a precision model. Semantic search finds the email that never uses your keyword; keyword search nails the exact phrase - together they miss less.
Can I use search operators like in an email client?
Yes. Operators such as from:, to:, after:, before:, kind:, label:, is:hot, and is:privileged can be typed straight into the query box, and the interface confirms which operators it applied so there is never a silent misread.
Will a query in English find documents written in another language?
Yes. Semantic search is language-agnostic by design, and the platform automatically translates queries for the keyword leg, so cross-language matters stay fully searchable from one query box.
Do newsletters and bulk mail drown out the evidence?
No. Bulk mail is screened out of ranking by default so it stops competing with real evidence - but nothing is deleted. Screened material remains in the index and is searchable on demand whenever you need to look at it.