eDiscovery AI vs Everlaw: how they compare in 2026

The models are the product, with nothing beneath them to license. Every capability sold is a classification or extraction produced by a language model: relevance coding against written instructions, privilege calls with named attorneys and privilege elements, early case intelligence, detection and extraction of personal information, and image comparison, filtering and facial recognition. No repository, processing engine, review platform or production tool is sold alongside them. The delivery design settles it. The product is a plugin that receives documents from Relativity, classifies them and writes results back to a mapped field. The platform of record belongs to someone else, and this vendor supplies the judgment applied to the documents. The vendor presents AI review as the successor to Boolean search and technology assisted review, not a feature added to either.

The platform covers upload, processing, review, production and trial preparation, and was sold before the generative layer arrived in 2024. The models drive a core layer on top of it. Coding Suggestions automates first pass review, Review Assistant and Writing Assistant summarize, extract and draft, and Deep Dive has taken more than 300 million documents and nearly 100,000 questions. Everlaw Omni, announced on 28 October 2026 for general availability in early 2027, adds an agent that runs Everlaw's tools and Case Memory, a library of extracted facts, and absorbs the Deep Dive screen. Predictive Coding and Clustering apply machine learning that predates the generative features. Everlaw presents itself as the evidence layer for legal AI, serving outside AI tools through its integrations, while the document review platform still runs without the generative layer.

Grounding is real and documented for each document, and a figure is published without the test set that would make it testable. The grounding is built into the output. Relevance restates the instruction it was given for each classification and explains why it categorized the document as it did, consistently across every document. Privilege goes further: it names the attorneys it identified in a separate field, sets out which elements of privilege are present, and drafts its reasoning as a privilege log entry. The published workflow ends in validation, with the fourth step being to review results and generate standard performance metrics. The vendor says the product consistently achieves 90 percent recall with excellent precision. No corpus, sample, matter type or method is given for the recall figure, precision is described but not quantified, and no failure mode is identified.

Everlaw AI cites source documents in its results so users can verify output, and Deep Dive answers draw on facts extracted from specific documents, retrieved from a vector database of the customer's own documents. Everlaw says grounding curtails hallucinations without removing them. Coding Suggestions gives tiered recommendations with a justification for every document, and the AI product page says it can exceed human review performance. No accuracy rate, recall or precision figure, test set or evaluation method from Everlaw itself is published. The 98.2% recall figure comes from one customer's layered workflow combining keyword culling, Coding Suggestions and Predictive Coding.

The review steps and the route back to human judgment are published as part of the workflow, short of any stated threshold. Oversight is built into the four published steps: the reviewer sets the criteria, the model classifies, and the fourth step is to review results to validate the classifications and generate standard performance metrics. That is a validation gate with a measurement attached. Every classification carries a restated instruction and an explanation, so a reviewer can audit a call rather than accept it, and privilege output is a draft, with the log entry field available to use or edit. The named attorney field exists so results can be validated. What is missing is the boundary. Nothing published states a confidence threshold, what the system does with an ambiguous document, an escalation path, or what share of a set a reviewer should sample. Nothing separates a confident classification from a doubtful one, which matters on privilege, where a single wrong call can waive it.

Control is the first of Everlaw's generative AI principles, beside Confidence and Transparency, Privacy and Security. Users can opt in or out of Everlaw AI, and an icon marks every generative feature. Everlaw AI cites the source documents in its results so users can check their work, and each statement in a draft is followed by the document meant to support it, one click away. The customer terms require the customer to evaluate output for accuracy, including through human review, before relying on it. Coding Suggestions gives tiered coding recommendations with a justification for every document, and in the Dinsmore & Shohl study attorneys validated its output against a 120 document ground truth set. Workflows, arriving in November 2026, can pause at checkpoints for a person's approval, and Omni keeps an audit trail and leaves the user to decide what is added to the work. No threshold at which a document must go to a person, no limit on what Coding Suggestions decides unaided, and no route for reversing a wrong AI output are published.

The published claim is about performance, not deployment: the product consistently achieves 90 percent recall with excellent precision, and reviews hundreds of thousands of documents in hours at a fraction of the time and cost of manual review. No matter, client, engagement, volume or method is attached, and nothing says whether the figure comes from a benchmark, a customer project or an internal test. No customer is named on the product pages. The product is listed on the Relativity App Hub and delivered as a Relativity plugin, a partner relationship rather than a customer. The vendor publishes case studies alongside fact sheets, whitepapers, videos and books.

Dinsmore & Shohl's study, published on 24 March 2026, describes a review of 26,000 contracts after a client's cyber incident. Partner Jennifer Mitchell and ediscovery director Peter Pepiton first culled about 20,000 drafts and duplicates. They then ran Coding Suggestions against attorney written criteria, validated it on a 120 document ground truth set, and finished with Predictive Coding. Recall rose from about 92% on the first Coding Suggestions pass to 98.2% after formal statistical validation. The home page adds stories from an unnamed Am Law 200 firm and an unnamed Fortune 500 company. Government agencies use Everlaw under its FedRAMP Moderate and GovRAMP authorizations.

Nothing is published on how client confidences are handled by a product built to take them in, and both published documents were read in full. The Privacy Notice, last updated 8 January 2025, applies only to information collected on or through the vendor's websites or other interactions with the vendor and its clients. It covers names, addresses, identifiers, browsing data and marketing preferences, and says nothing about customer documents submitted for classification. The Terms and Conditions, last updated 23 May 2024, are a generic website terms of use with no data provisions. So nothing published addresses training on client documents, retention of a reviewed collection, segregation between matters or clients, or what any model provider may keep. The product's main function is to identify privileged material so it can be withheld, yet the vendor publishes nothing on how it protects the privilege of the documents it reads.

The customer terms of 30 October 2026 bar Everlaw from using Customer Data, which includes inputs and outputs, to train or fine tune a generative AI model, and inputs and outputs remain case materials the customer owns. The enterprise language models follow a zero data retention policy, using data only to generate a response and deleting it on completion. Data is not shared between customers, and the models hold no persistent knowledge of a customer's case that a data injection could extract. The vector database behind Deep Dive keeps numerical embeddings created from customer documents, which retrieval needs, and Omni's Case Memory keeps facts extracted from them. Everlaw describes its environments as built for sensitive case data, privileged communications and personal information. No retention period is stated for the embeddings or Case Memory.

Nothing published addresses the advice line for a product that makes privilege calls. The audience is clear: the product is sold to legal teams and eDiscovery service providers and delivered inside Relativity, so there is no consumer confusion to resolve. Nothing else is addressed. No statement is published that the vendor is not a law firm, that a classification is not a legal determination, or that a privilege call needs attorney review before it is used to withhold a document. Nothing covers supervision, competence or the reviewer's own professional duties, and the two published documents, both read in full, contain no disclaimer on the point. That matters because a privilege designation is a legal conclusion with a waiver consequence, and the published workflow has the model make it first. The validation step does put a person after the machine, but that is a control, not a professional responsibility position.

The buyers are law firms, corporate legal departments and government legal teams, with no consumer product. The customer terms require the customer to review output for accuracy before relying on it and disclaim any warranty on AI output. Everlaw is a document review and case building platform rather than an advice tool, and it publishes nothing on advice versus tooling, on competence or supervision duties, or on jurisdiction limits.

No governance position is published for a system whose output affects legal outcomes. There is no responsible AI page, principles, named framework, certification, accountable person, description of what is evaluated before a model changes, or testing regime beyond the project metrics a customer generates itself. Neither published document mentions AI. Bias is unaddressed, and it is a live issue here. The vendor ships facial recognition as a named feature, alongside detection of personal information, applied to litigation collections that contain images of real people. Facial recognition is the AI capability with the most documented demographic performance gaps, and nothing published addresses accuracy across populations, thresholds, or what a false match in document review would mean. The relevance and privilege classifiers raise a quieter version of the question, since a model reading correspondence to infer legal significance can perform unevenly across writing styles, languages and formats.

An AI Governance Framework rests on three generative AI principles: Control, Confidence, and Transparency, Privacy and Security. Third party technologies are evaluated from technical, legal, privacy and security perspectives before adoption, and accuracy assessments inform which language models Everlaw uses. Users can opt in or out of Everlaw AI, and the framework sets out protections against data injection and exfiltration. On bias, the framework says Everlaw cannot manage bias inside third party foundation models and does not use prompt based mitigation, because that could mask bias in the case documents themselves. No accountable owner of AI governance is named and no testing results are published.

Nothing is published on retention, deletion or access for a system that holds client documents, and both documents that would carry it were read in full. The Privacy Notice's security section is one paragraph. It says the vendor takes a number of steps to protect against loss and misuse of information under its control, warns that no internet transmission is completely secure, says information is sent at the user's own risk, and disclaims responsibility for illegal acts such as hacking by third parties. No encryption, access control, monitoring or incident practice is described. Its retention section sets criteria rather than periods and covers personal information collected through the website, not a reviewed document collection. There is no subprocessor list, security page, trust center or data processing addendum, and the full site navigation has no security page. The Terms and Conditions contain no data provisions. For a product that takes in whole litigation collections, that is the entire published record.

Under a zero data retention policy for the language models, data sent for processing is used only to generate a response and deleted on completion. Everlaw itself keeps customer data only as long as the customer stores it on the platform. A Pinecone vector database stores embeddings of customer documents to power Deep Dive, and Omni's Case Memory adds a store of facts extracted from them. The data processing addendum, dated 30 October 2026, commits to notice of a verified security incident without undue delay, with information as it becomes known and steps to contain and remediate. Subprocessors are listed with function and country, and additions or replacements are notified at least 10 days ahead. After termination the customer has a 30 day grace period to export data, and Everlaw then deletes it within a commercially reasonable timeframe. Customers choose the AWS instance where data is stored and processed. No period is stated for deletion after termination, for the embeddings behind Deep Dive or for Case Memory.

Nothing is published on who bears the loss when the system is wrong. The Terms and Conditions were read in full. They run to about a dozen short clauses on content ownership, access to external resources, acceptable use, no waiver, service interruption, reselling, intellectual property, changes to terms, assignment, contacts and severability. They contain no limitation of liability, warranty disclaimer, indemnity or cap of any kind, so nothing is allocated in either direction. Nothing addresses a relevant document coded nonresponsive and not produced, or a privileged document classified as not privileged and produced, which is the costliest failure in this workflow and the one the product exists to prevent. No warranty attaches to the published 90 percent recall figure, no insurance position appears, and no customer agreement is published.

An indemnity from Everlaw against third party claims that the service infringes intellectual property rights sits in its Customer Terms of Service, last updated 30 October 2026. Claims that Everlaw AI output infringes are excluded, along with customer negligence, breach, alteration, combination with other products and beta services. Each party's liability is capped at the fees paid or payable in the 12 months before the liability arises, and indirect and consequential damages are excluded, with a $100 cap for beta services. Everlaw warrants compliance with applicable law, material conformity with its documentation and enterprise security software against harmful code. It makes no warranty on the accuracy, completeness or reliability of AI output, and the customer must review output before relying on it. Section 13.3 commits Everlaw to carry standard and customary insurance at its own cost.

One integration, documented at the level of what actually moves. The product is a Relativity plugin listed on the Relativity App Hub, a reviewed partner channel rather than a self declared compatibility claim. The workflow is described: the reviewer selects documents in Relativity, supplies relevance, issue, privilege and category instructions, submits the documents through the plugin, and receives results written back to a mapped field. The vendor says the only setup needed is the prompt instructions and the field to map results to, and that the same instructions can run on more documents without new indexing or model training. Breadth and documentation are missing. No other platform is named, so a firm on Everlaw, Reveal, DISCO or Nuix has nothing published to work from. No API, developer documentation or connector specification is published, and nothing describes authentication, volume limits or how results are versioned.

Connections reach the AI tools legal teams already use, with Claude, ChatGPT Enterprise, Google Gemini, Microsoft Copilot, Harvey, Legora and Thomson Reuters named as integration partners, each described in a line. Claude can retrieve and analyze case materials directly, and ChatGPT Enterprise uses Everlaw as an evidence layer. API and MCP connectors let firms and partners build custom tools on Everlaw as the evidentiary system of record. Cloud connectors collect data straight from Microsoft 365, Zoom and Google Workspace. No public page sets out which objects each integration moves, in which direction, or what a firm must configure.

Cloud delivery is implied, and neither the tenancy model nor the region is stated. Documents go from a customer's Relativity workspace to eDiscovery AI for classification, and results return to a mapped field, so processing happens on the vendor's side through a hosted service. Nothing more is published. No cloud provider, region or data center is named, no residency option is offered, no tenancy model is described, and nothing says whether one customer's collection is isolated from another's. Nothing separates where documents are held during processing from where inference runs, and no on premises or in tenant option is mentioned. The only geographic statement is in the Privacy Notice: the vendor operates in various countries, including the United States and countries in the EU and EEA, and personal information may be accessed from outside the United States. That concerns website personal information, not document collections.

Data is stored and processed in the Amazon Web Services instance the customer chooses: United States, United Kingdom, EU in Germany, Canada or Australia. AI processing follows the instance in the United States, the United Kingdom, the EU and Canada, with Google and OpenAI processing in the EU for UK and EU customers and OpenAI in Canada for the Canadian instance, under the same zero retention and no training terms. The subprocessor list gives no Australian processing location for Google or OpenAI. Government agencies run on Everlaw's FedRAMP Moderate and GovRAMP authorized environments, and Everlaw for Word is offered outside GovCloud. The transparency report says Everlaw is subject to United States law and that data may also be subject to the laws of the chosen instance's country. No tenancy model is described.

No independent security attestation is published. The site navigation and footer show Solutions, Partners, Thought Leadership, Company and Contact, with Privacy Policy and Terms and Conditions in the footer. There is no security page, trust center or compliance page. The Privacy item in the Solutions menu is a product for detecting and extracting personal information, not a policy page. No certification of any kind is claimed: no SOC 2, ISO 27001 or HIPAA, and no badge, auditor, report or request route. No penetration testing or vulnerability disclosure program is described. The only security statement is the Privacy Notice paragraph, which describes no control and disclaims responsibility for third party hacking. There is nothing to request.

Everlaw holds FedRAMP Moderate and GovRAMP Moderate authorizations, SOC 2 Type 2 and SOC 3 reports, ISO 27001:2022, ISO 27017 and ISO 27018 certification and Cyber Essentials Plus, and says it supports its customers' compliance with GDPR and CCPA. Its security and compliance portal at trust.everlaw.com lists the SOC 2 Type II report with a 2026 bridge letter and a penetration test executive summary. It also lists a HIPAA HITECH assessment, a CAIQ dated May 2026, a SIG Lite and the ISO statement of applicability. Reports open through an access request on the portal. The portal also carries an architecture diagram, the AI governance framework and a 2026 disaster recovery test summary.

The vendor refers to advanced AI without identifying anything underneath it. The language across the site stays at the level of capability and is consistent: advanced AI technology, powerful and efficient AI technology, generative AI, and AI review technology that understands grammar and context. No model is named, no version is given, no provider is identified, no hosting arrangement for inference is stated, and nothing commits to notifying a customer when any of it changes. There is no subprocessor list on any surface, and neither published document names a processor. One architectural fact is disclosed: the vendor states that no additional indexing or model training is required to run the same instructions against a further document set. The product thus works by instruction rather than by fine tuning on the customer's collection, a statement about method rather than supply. Documents leave the customer's Relativity workspace and go to the vendor for classification, so a litigation collection reaches a model no one has named.

Amazon Web Services and Google appear on the subprocessor list, updated 2 October 2026, as cloud and AI service providers, with OpenAI as an AI service provider and Pinecone for the vector database behind Deep Dive. It gives countries of processing by instance, with Google and OpenAI processing in the EU for UK and EU customers and OpenAI in Canada for the Canadian instance, while a June 2024 FAQ still says OpenAI has servers only in the United States. The integrations page says Google Gemini powers retrieval and context engineering. The data processing addendum commits to notice at least 10 days before a subprocessor is added or replaced, with 15 days to object. The AI governance framework says the language models follow zero data retention terms and that accuracy assessments inform which models Everlaw uses. No model or version is named for any feature, including Omni.

No pricing information is published at any level, including the unit of charge. There is no pricing page. No plan, tier, rate, band, minimum or term is stated, and nothing says whether the product is charged per document, per gigabyte, per matter, per classification run or by subscription, the first question a litigation support manager would ask. The only commercial route is a demo booking, repeated on every page, with a phone number and a support address as the only contacts. The vendor's central commercial claim is that review is done at a fraction of the time and cost of manual review, and no figure lets a buyer test that against what contract attorney review would cost.

What the platform includes, rather than what it costs, fills the pricing page. Cloud connectors, processing at one million documents an hour, transcription, redactions, productions, Predictive Coding, Clustering, timelines, transcript analysis, onboarding, training, standard data migration and unlimited user licenses are included. Coding Suggestions, Deep Dive, machine translation and custom extractions run on credits, with single document use free of credits. An intake portal with case management and compliance reporting is an add on. No figure, tier or unit of charge for the platform is published, and pricing runs through a demo request.

Coverage is described by capability rather than by who the product serves, and the breadth is claimed without evidence. What is published is a product list: four suites covering early case intelligence, review, privacy and multimedia, each with named parts. The buyer is described only as legal professionals and legal teams, a Partners page shows eDiscovery service providers as a channel, and the Relativity plugin implies a firm or provider already on that platform. No practice area is named, no matter type is called suited or unsuited, no firm segment or size is addressed, and no line is drawn between what a law firm, corporate legal department or service provider would use. There is no jurisdictional statement at all. That matters on a product making privilege calls, since privilege doctrine and the treatment of in house counsel communications differ between federal, state and foreign practice, and nothing says which the privilege classifier is built for.

Everlaw sells to law firms, corporate legal departments and government agencies at federal, state and local level, and says its customers include 93 of the Am Law 200, all 50 state attorneys general and Fortune 100 legal departments. Public sector use cases include litigation, investigations and Freedom of Information Act requests. Private sector use cases include document review, early case assessment, legal holds, internal investigations, data privacy, depositions and trial preparation. Its practice scope is ediscovery, investigation and litigation from data upload through production to trial, with no claim to transactional or advisory work. Regional instances run in the United States, United Kingdom, Germany, Canada and Australia. Segmentation by firm size, industry breakouts and any statement of what the platform is not built for are not published.