DISCO Advanced Research

DISCO Advanced Research is an agentic AI application for legal fact investigation that reached general availability in August 2026 inside the DISCO Ediscovery platform. It is built for litigators and eDiscovery teams who need to interrogate a matter's full document corpus, and it replaces iterative manual searching with autonomous multi-step reasoning that produces memos, deposition outlines and timelines in minutes.

DISCO Advanced Research is an agentic AI application for legal fact investigation that reached general availability in August 2026, following an early access period that began in summer 2026. It is built by CS Disco (NYSE: LAW), the Austin-based cloud eDiscovery vendor founded in 2013, and runs inside the DISCO Ediscovery platform alongside the company's Cecilia AI suite. Where conventional eDiscovery search retrieves documents matching a query, Advanced Research applies autonomous multi-step reasoning: it decomposes a complex litigation question, runs both keyword and semantic searches across the matter's document corpus, reviews its own findings to identify what is missing, and issues further searches until it can produce a well-supported answer — then exposes the decision trail so a litigator can audit how each conclusion was reached, which matters in a domain where an unexplained AI answer is unusable. Outputs are practitioner-shaped rather than generic: memos, deposition outlines, chronological timelines and review-ready result sets, delivered in minutes, with granular connections carried across subsequent queries. DISCO reported that during early access, customer engagement more than doubled, with those users asking roughly twice as many questions on average as non-early-access users. Commercially it arrives inside the all-inclusive platform DISCO announced on 25 February 2026, which collapsed separately priced modules — ediscovery, Cecilia AI, and Deposition Management and Timelines, formerly sold as Case Builder — into a single transparent per-gigabyte rate for processed data with no ingest fees; the agentic capability is included at no additional cost rather than sold as an AI upcharge, which is the notable contrast with competitors metering AI separately. DISCO reported 1,549 customers, 330 large customers and 98% dollar-based net retention as of 31 December 2025, on trailing-twelve-month revenue of $162 million as of 31 March 2026.

Category

Industry & Government

Pricing

Usage-based, Contact for pricing

Target Market

General Counsel, Litigation Partners, eDiscovery Managers, Legal Operations Leaders, CIOs

Deployment

Cloud-only

Founded

2013

Headquarters

Austin, Texas, United States

Customers

1,549 customers including 330 large customers as of 31 December 2025

• ✓
  Autonomous multi-step agentic reasoning

  Decomposes a complex litigation question into a chain of searches and analyses rather than returning one ranked result list per query

• ✓
  Self-review and gap detection

  The system reviews its own findings, determines what information is missing, and executes additional searches before answering

• ✓
  Transparent decision trail

  Exposes how each conclusion was reached, which is what makes the output defensible enough for a litigator to rely on

• ✓
  Combined keyword and semantic matching

  Matches on both literal terms and meaning, so evidence phrased differently from the query is still surfaced

• ✓
  Practitioner-shaped outputs

  Produces memos, deposition outlines, timelines and review-ready result sets in minutes instead of raw document hits requiring manual synthesis

• ✓
  Included in all-inclusive per-GB pricing

  Bundled at no additional charge with the February 2026 platform, so AI usage does not create a separate unpredictable line item on a matter budget

• ✓
  Data isolation from model training

  DISCO states its AI features never use customer files for LLM training, which is a precondition for privileged material

• •
  Early case assessment on a new matter

  Counsel asks broad factual questions of the full corpus in the first days of a matter to shape strategy before large-scale review begins

• •
  Deposition outline generation

  The agent assembles a witness-specific outline with supporting documents drawn from the reviewed set, compressing days of preparation into minutes

• •
  Chronology and timeline construction

  Builds a dated sequence of events across email, chat and documents so the team can see how a dispute actually unfolded

• •
  Iterative fact investigation across queries

  Granular connections carry across subsequent questions, so each answer sharpens the next rather than restarting the search from zero

• •
  Responding to opposing counsel's narrative

  A litigator tests a factual assertion against the evidence quickly enough to respond within the pace of active litigation

DISCO moved to a single all-inclusive commercial model on 25 February 2026: one transparent flat per-gigabyte price for processed data, no ingest fees, and every module — ediscovery, Cecilia AI, deposition management and timelines, and the Advanced Research agentic application — included rather than priced separately. The headline rate is not published, so the per-GB number comes from a quote, but the structure is designed for like-for-like comparison against competitors that meter AI features separately, and G2 reviewers repeatedly cite billing transparency and the absence of hidden costs as a strength. The corresponding complaint is absolute cost: reviewers note it is hard to justify for some clients and smaller matters, so the per-GB rate is the number to negotiate hard on.

This page was written from 6 sources, 4 on domains other than csdisco.com.

1. 1.stocktitan.net — disco launches advanced research for deep agentic ai powered
2. 2.csdisco.com — disco announces all inclusive platform for ediscoveryvendor
3. 3.g2.com — reviews
4. 4.aivortex.io — best e discovery ai tools
5. 5.csdisco.com — securityvendor
6. 6.stocktitan.net — ars cs disco inc sec filing 5176e78ca160