Executive Summary
Discovery is where litigation costs are made. The platform decides how much of that cost is people reading documents.
Electronic discovery is the process of finding, preserving, collecting, reviewing and producing electronically stored information (ESI) for a legal matter. Microsoft says Microsoft Purview eDiscovery is used to identify, review and manage content in Microsoft 365 services. The work runs from the legal hold that stops deletion, through collection and processing, to review for relevance and privilege, and production to the other side. The contract intelligence guide covers transactional legal work; this guide covers litigation and investigations.
AI has changed review, the most expensive step. Platforms now offer generative AI review alongside earlier technology-assisted review. The approaches in this guide differ by who runs the work and at what scale: review platforms with partner ecosystems, cloud eDiscovery for law firms and corporate legal teams, self-service tools, managed review services, and eDiscovery inside the productivity suite.
Why eDiscovery Is a Separate Purchase
Courts expect parties to preserve and produce relevant information, and they can sanction failures. In US federal courts, the Federal Rules of Civil Procedure set the standard. The rule on failure to preserve electronically stored information applies when such information that should have been preserved in the anticipation or conduct of litigation is lost because a party failed to take reasonable steps to preserve it, and it cannot be restored or replaced through additional discovery. The cost of review also limits what can be asked for. Parties may obtain discovery regarding any nonprivileged matter that is relevant to any party's claim or defense and proportional to the needs of the case.
The industry's reference model has just changed. EDRM says EDRM 2.0, released September 1, 2026, is the first substantive update to the Electronic Discovery Reference Model since it incorporated the full Information Governance Reference Model. EDRM defines preservation as safeguarding relevant ESI and documents against alteration or destruction. Use the model to check that a platform, or a combination of platforms, covers every stage you need.
Platform, Self-Service Tool, or Managed Service
Legal teams choose among a full platform run by their own staff or a provider, a self-service tool for smaller matters, a managed service that runs the work, or eDiscovery built into the productivity suite. Some organizations use more than one: built-in tools for holds and early collection, a platform or service for large reviews.
Built-in tools depend on licensing. Microsoft says access to specific eDiscovery or premium eDiscovery features depends on the licensing and subscriptions of the organization. Self-service tools aim at speed and simplicity. Logikcull says thousands of legal teams use it to handle eDiscovery, FOIA requests and legal holds.
| Model | What you get | What you take on |
|---|---|---|
| Review platform | Processing, review and production at scale | Staff or partners to run it |
| Cloud eDiscovery | Holds through production in one cloud product | Fit for very large or unusual matters |
| Self-service tool | Fast start with simple pricing | Limits on scale and customization |
| Managed service | Experts and reviewers who run the work | Less direct control and service fees |
| Suite-native eDiscovery | Holds and searches inside the productivity suite | Coverage limited to that suite's data |
How do you evaluate AI eDiscovery Platforms?
Evaluate eDiscovery platforms on a closed matter from your own files. The capabilities that decide results are legal holds, collection from your data sources, processing, search and analytics, AI review, privilege handling, production, and security.
Holds and preservation come first. DISCO describes a single, cloud-based platform to preserve data, notify custodians and track holds with a defensible audit trail. Everlaw describes preserving data at the source, automating custodian tracking and maintaining a full audit trail. AI review is the newest layer. DISCO says Auto Review with Tag Tuner lets teams run AI-powered document reviews. Relativity says external AI applications can be integrated with Relativity aiR.
| Capability | What it does | Buyer translation |
|---|---|---|
| Legal holds | Notifies custodians and preserves data | Ask how holds reach each data source. |
| Collection | Gathers data from email, chat and files | Match to the sources you use. |
| Processing | Extracts text and metadata | Ask about volume limits and speed. |
| Search and analytics | Finds key documents and patterns | Test on a closed matter. |
| AI review | Classifies documents for relevance and privilege | Ask how results are validated. |
| Privilege handling | Identifies and logs privileged material | Ask how privilege logs are produced. |
| Production | Prepares documents for the other side | Ask which formats are supported. |
| Security and hosting | Protects sensitive case data | Ask where data is hosted and who can access it. |
Vendors in AI eDiscovery Platforms
Representative vendors include DISCO, Everlaw, Lighthouse, Logikcull, Microsoft and Relativity. The list is alphabetical, is not a ranking, and is not for sale. How we choose vendors
The approaches below differ by who runs the work and the size of matters they suit. Several vendors offer software and services, so each approach describes where a provider is strongest.
Platforms are opening to outside AI and data. Everlaw says customers can connect AI tools, agents and workflows to evidence in Everlaw's system of record. Reveal says the full Reveal platform can be hosted however and wherever the customer wants. The market has also consolidated: Zapproved's site states that Zapproved has been acquired by Exterro. Ask any vendor that has changed hands about its roadmap and support.
| Vendor | Approach | Where it fits |
|---|---|---|
| DISCO | AI review and holds | Cloud eDiscovery with AI review and holds |
| Everlaw | Cloud eDiscovery | Evidence system of record from hold to trial |
| Lighthouse | Managed review services | Managed eDiscovery with AI-driven insight |
| Relativity | Review platforms | Review platform with aiR and outside AI integrations |
| Logikcull | Self-service eDiscovery | Self-service discovery with published plan types |
| Microsoft | Suite-native eDiscovery | eDiscovery for content in Microsoft 365 |
The approaches were set before one provider was chosen for each, and a row here is never for sale. Close to twenty providers sell eDiscovery software or services. Providers, named here or not, can make their own case in the Spotlight below.
How much should you budget for AI eDiscovery Platforms?
eDiscovery is priced by data volume, matters, users and services, and AI review adds its own charges. Logikcull offers storage-based plans, which include reusable blocks of data storage, and matter-based plans, which include a certain number of matters and charge for data uploaded on a one-time basis. Logikcull says its pay-as-you-go plans are offered monthly with no commitment, priced on the amount of data stored with a minimum threshold of 10 GB.
Review hours can be the largest cost in a matter, so the value of AI review shows up in fewer documents read by people. Model it on a recent matter: data volume, documents reviewed, review hours and rates.
| Meter | You are charged for | Grows with | Where it goes wrong |
|---|---|---|---|
| Data volume | Gigabytes processed or hosted | Matter size and duration | Hosting charges on closed matters |
| Matters | Cases or projects | Matter count | Small matters priced like large ones |
| Users | Reviewers and administrators | Team size | Paying for contract reviewers by seat |
| AI review | Documents classified by AI | Volume reviewed | AI charges added to every document |
| Services | Processing, review and consulting | Scope | Hourly services without caps |
How long does implementation take for AI eDiscovery Platforms?
eDiscovery programs depend on knowing where data lives and on a repeatable, documented process. Start with holds, then build the review workflow.
Map data sources and custodians. Set up legal hold notices and preservation for each source.
Connect collection from priority sources. Test processing on a closed matter.
Configure review, privilege and production. Validate AI review against known outcomes and document the method.
Roll out to new matters. Track cost per matter, review hours and hosting.
What should you ask vendors about AI eDiscovery Platforms?
Ask each provider to answer on your data sources and a closed matter.
-
Do you run large, complex matters often?Yes Shortlist review platforms and AI review products, or managed services.No Continue.
-
Is your legal team small with many small matters?Yes Shortlist self-service eDiscovery.No Continue.
-
Is most of your data in one productivity suite?Yes Start with suite-native holds and search, and add a platform for review.No A cloud eDiscovery product fits.