DISCO vs Relativity: how they compare in 2026
DISCO and Relativity are cloud ediscovery platforms bought by the same litigation teams. Relativity's trust site names ISO/IEC 27001:2022, ISO/IEC 27018:2019, FedRAMP Moderate authorization, IRAP at Protected, HIPAA, SOC 2 Type II, a public SOC 3 report and a Cloud Security Alliance questionnaire. Customer Lockbox, on by default, keeps Relativity's own administrators out of a customer workspace unless access is granted, and a Redgrave LLP study reports aiR Review's recall against active learning. DISCO publishes its terms, data processing addendum and subprocessor list, naming Writer and OpenAI as its language model providers. It bills per GB of processed data with unlimited Cecilia AI included, and Auto Review classifies up to 32,000 documents an hour with an explanation for every tag and recall and precision tracked in the product.
At a glance
All 15 axes, side by side
The same grid applied to every vendor in the index, graded from public sources. Hover a grade to see what the letter means on that axis.
AI Centrality
How much of the product is actually AI. Whether the machine learning is the mechanism the buyer is paying for or a feature layered onto conventional software, and whether the vendor is specific about which is which.
The models drive a core layer on a platform that works without them. DISCO Ediscovery covers upload, filtering, review and production, and Cecilia AI arrived in 2023, about a decade after the company was founded. Cecilia Q&A, single document Q&A, summaries, definitions and Auto Review's first pass classification run on language models. DISCO now sells unlimited Cecilia AI inside its per GB rate, so the generative features come with every matter on the platform.
The platform began as an ediscovery system and still runs legal hold, collection, processing, review and production without generative AI. aiR Review, Privilege, Case Strategy, Data Breach Response and Assist sit on that base and drive its review and analysis work. Older model based features remain in use, among them the Prioritized Review classifier that learns from reviewers' coding, conceptual analytics on latent semantic indexing and sentiment analysis. Every AI capability is included with a committed data volume, and the platform is now sold under the aiR name.
Citation Accuracy and Hallucination Disclosure
Whether the vendor publishes measured accuracy on citations and assertions, grounds output to primary sources, and says plainly what its system does when it does not know. Legal has a documented public record of fabricated citations reaching filed briefs, so an untested claim of accuracy is not evidence.
Cecilia Q&A answers only from the customer's evidence, with citations, and Auto Review gives an explanation for every tagging decision so a reviewer sees why each classification was made. Auto Review tracks recall and precision inside the product, and DISCO's AI consulting team advises on sampling and validation. DISCO says Auto Review delivers precision and recall 10 to 20 percent above human reviewers. No test set, date, evaluator or method accompanies that figure, so an outsider cannot check or reproduce it.
A study by Redgrave LLP compared aiR Review with a managed first pass review using active learning on about 45,000 documents from a public data set. Responsiveness turned on federal pharmaceutical compliance rather than topic, and a subject matter expert set the ground truth through a blind review of a random sample. aiR Review reached 88 percent recall against 64 percent, with an elusion rate of 1 percent against 3 percent. It also flagged more documents, at 29 percent precision against 39 percent, a trade off the study sets out. aiR Review used about 18 hours of attorney time against about 1,123, and the full report gives the method and statistics. Each aiR Review call cites the document passages it relies on.
Autonomy and Oversight Model
What the system decides on its own, what a lawyer must approve, and whether the vendor documents where the review point sits. A tool that drafts under review and a tool that files without one are different products and different risks.
Auto Review suggests tags with an explanation for each, and the team moves to quality control on those suggestions. Customers choose whether to run Auto Review themselves, refining tags with Tag Tuner, or to have DISCO's Managed Review team run the project, including privilege review. Recall and precision are tracked in the product so a team can validate the review before relying on it. Cecilia answers cite the evidence they draw on. No threshold at which a document must go to a person, and no limit on what Auto Review decides unaided, is published.
aiR Review explains why each document matters and cites the passages behind each call. aiR for Privilege predicts whether documents may be privileged, identifies the people involved and drafts a privilege log description. Prioritized Review learns from reviewers' coding decisions to rank what people see next and can run beside aiR Review. In the Redgrave LLP study, a subject matter expert reviewed a blind sample to set the ground truth before revisiting the documents where he and aiR disagreed. No threshold is published at which a document must go to a person, and no commitment covers what happens when an aiR call is wrong.
Operational and Outcome Evidence
Named, dated evidence that the product works in production at real firms or legal departments. Case studies with figures and identified customers count. Unattributed testimonials and launch announcements do not.
CS Disco's annual filing, a regulated document, reports 1,549 customers including 330 large customers and a 98 percent dollar based net retention rate at 31 December 2025. Named customer material is published too. A Perkins Coie associate is quoted on Auto Review, and a story on Rhoades McKee describes DISCO's professional services on an inherited multi jurisdiction case. Auto Review is said to deliver precision and recall 10 to 20 percent above human reviewers, with no test set, date or method given. No named customer is paired with dated figures and a method a reader can assess.
JND, a Relativity partner, ran aiR Review on a corporate client's production and reports more than 750 hours and $85,000 saved, with project time cut by nearly 90%. JND checks aiR's calls on a random sample of about 400 documents and computes recall and precision. Microsoft's legal department reports average recall above 96% and precision of 71% on an antitrust matter, review costs down 45% to 60% and $1.6 million in projected annual savings. Lowenstein Sandler and Baker McKenzie are quoted on claiR. The customer stories carry no date.
Privilege and Confidentiality Posture
How client confidences are handled: attorney client privilege and work product treatment, segregation of one client matter from another, whether client data trains any model, and what the vendor commits to in writing rather than in marketing.
The language models used in Cecilia never train on or learn from customer data and usage, DISCO says, and do not retain customer data after a query. The models behind Auto Review are prohibited from training on or retaining customer data after processing, and model vendors must hold SOC 2 Type 2 or ISO 27001 certification. Data is encrypted in transit and at rest. The data processing addendum binds DISCO to confidentiality, breach notice and deletion or return of personal data on termination, and notice of subpoenas or other orders seeking personal data. No position on privilege or work product as such, and no separation between users or matters inside a customer, is published.
Discrete permissions control what each user reaches, and Lockbox bars administrators, Relativity staff among them, from customer data without the customer's say. Data is encrypted at rest and in transit, with customer managed keys available. Customer data processed by aiR does not train or improve third party models, and AI processing runs in the published location for each instance region. Nothing addresses training of Relativity's own models, retention periods or the handling of privileged material inside aiR.
UPL and Professional Responsibility Posture
Whether the vendor is clear that it supplies a tool rather than legal advice, who its audience is, and how it addresses unauthorized practice of law, competence and supervision duties, and jurisdiction limits. ABA Formal Opinion 512 is the reference point. Where the advice line is not the duty a product raises, the axis is read through the nearest professional duty it does raise: judicial conduct rules and the reviewing duty for products sold only to courts, and the duty to bill for time actually spent for products that draft time entries.
The buyers are law firms, corporate legal teams and government legal teams, with no consumer product, and DISCO analyzes collected evidence rather than giving advice. Auto Review instructions can be written with DISCO's AI consulting team, and its Managed Review team can run a first pass review and handle privilege review, which places vendor staff inside the customer's review method. Nothing published addresses advice versus tooling, the supervising lawyer's responsibility for a review DISCO's team runs, competence duties or jurisdiction limits.
Law firms, corporate legal departments, government agencies and legal service providers use the platform, and no consumer product is offered. aiR reviews and analyzes a matter's documents rather than advising clients on the law. No position is published on advice versus tooling, on competence and supervision duties or on jurisdiction limits.
AI Governance and Bias Disclosure
Published governance over model behavior: who owns it inside the vendor, what is tested before release, and what is disclosed about disparate output across matter types, parties, or populations.
Every Auto Review tag suggestion carries an explanation, and Auto Review tracks recall and precision inside the product so a team can validate a review. DISCO's AI consulting team advises on sampling, validation and defensibility. The annual filing discusses AI related operational and regulatory risk beside cybersecurity, privacy compliance and reliance on third party model and cloud providers. No AI governance framework, AI principles, ISO 42001 certification, named owner of model governance or prerelease testing regime is published. Nothing addresses uneven output across matter types, parties or populations, for a product that classifies up to 32,000 documents an hour.
Under its AI Principles, Relativity commits to building AI for defined legal data tasks, to explaining how its models are trained and what each one looks at, and to testing models for bias. Where bias appears, the principles commit it to pause, consider mitigations, document its decisions and tell customers each model's appropriate use. The principles also state plainly that AI systems, Relativity's included, can give wrong results or contribute to biased decisions. aiR systems are updated regularly to improve quality, accuracy and consistency, so their output can change over time. No AI management certification, named accountable owner or bias test result appears.
AI Safety and Data Stewardship
Retention, deletion, access control, and what happens to prompts and documents after they are processed. Whether the vendor states its subprocessors and its incident practice, or leaves the buyer to assume.
DISCO's data processing addendum commits to notice of a personal data breach without undue delay, with assistance afterward, and on termination DISCO deletes or returns customer personal data or provides a way to do so. Its subprocessor list names each provider with function and country, and customers have 14 days to object after the list is updated. The language models behind Cecilia and Auto Review do not train on or retain customer data after processing, and every model vendor must hold SOC 2 Type 2 or ISO 27001 certification. Data is encrypted in transit and at rest on AWS. No period is stated for keeping case documents, Cecilia prompts or generated output, and retention under DISCO's own document retention policies is reserved.
By default, Relativity encrypts data at rest and in transit, and customer managed keys give the customer control of its encryption key. Lockbox keeps administrators, Relativity's own included, out of customer data unless the customer grants access, and customers can audit and revoke that access in real time. Security Center, built by the Calder7 security team, maps logins, flags users without two factor authentication and sends alerts on unusual events. No retention period, deletion schedule for prompts and aiR output, or subprocessor list is published.
AI Liability and Recourse
What the vendor stands behind contractually when its output is wrong. Indemnities, caps, carve outs, insurance, and whether any of it is published or only reachable through a negotiated agreement.
Third party claims that the technology offerings or documentation infringe United States intellectual property rights are covered by an indemnity in the terms and conditions, with the usual exclusions. DISCO warrants that each technology offering conforms in all material respects to its documentation, and that professional services are performed in a professional and workmanlike manner, with correction or termination as the remedy. Each party's liability is capped at the fees paid or payable under the order in the 12 months before the claim. A separate data protection cap of twice the general cap covers breaches of data protection obligations. Fraud, wilful misconduct, negligence causing death or injury and indemnification obligations sit outside the caps. The AI supplementary terms exclude AI services from any service level agreement, and no insurance position is published.
No customer agreement, master terms or service level terms are published, so indemnity, liability caps, warranties and insurance for aiR output cannot be read before contracting. AI processing locations and aiR Unit fees are set by each customer's agreement, and Relativity may rate limit or queue AI features at its discretion. aiR for Privilege predicts privilege and drafts log descriptions, and nothing published allocates the loss if a privileged document is produced on its call.
Practice Systems Integration Depth
How deeply the product reaches into the systems legal work already lives in: document management such as iManage and NetDocuments, Word and Outlook, contract lifecycle management, matter management, e-billing, and court filing systems.
The Hold module connects to a customer's systems, and the high speed uploader moves enterprise data straight into the platform. DISCO publishes guidance on Google Workspace collection through Google Vault and on migrating data from other ediscovery providers. A DISCO API is published for connecting the platform to a firm's other systems, and the subprocessor list names Box and Dropbox as file transfer tools. The consolidated integrations page now returns a not found page, and no documentation per integration sets out what moves, in which direction or what an administrator configures.
Collection connectors reach Microsoft 365, Google Workspace, Slack, Claude Enterprise, ChatGPT Enterprise and Gemini, with new data sources added regularly. An MCP server lets assistants such as Claude, Gemini, Copilot and ChatGPT set up workspaces, organize case data and manage matters in natural language. Open APIs and the App Hub extend the platform at no extra license cost, and Gavel brings drafting into Microsoft Word. What each connector moves, in which direction, and what an administrator configures are not described.
Deployment Model and Data Residency
Where the software runs and where the data sits. Multi tenant cloud, single tenant, private deployment, on premises, and whether region of residence is a published option or an enterprise conversation.
Cloud hosting on AWS spans the United States, Ireland and Canada, according to the subprocessor list. Microsoft Azure and Google appear as further hosting providers in the United States, and OpenAI processes in the United States and the EU. The platform was designed to operate in the cloud, with AWS used to minimize data footprint and strengthen privacy controls. The data processing addendum sets out transfer mechanisms for data leaving the EEA, the UK and Switzerland. No customer choice of region, tenancy model or private deployment is described.
Instance regions for Relativity aiR span the United States, Canada, Brazil, the UK, Ireland, France, Germany, the Netherlands, Switzerland, the UAE, South Africa, India, Australia, Japan, Hong Kong, Singapore and South Korea. All of them run on Microsoft Azure. Generative AI processing can run outside an instance's own region, and Relativity publishes where it runs for each one. French, German, Irish, Dutch, Brazilian and South African instances process within the EU Data Boundary, UK and UAE instances in the UK, and Hong Kong, Japanese, Singaporean and Korean instances in Japan. The United States, Canada, Switzerland, Australia and India process in country. aiR Government runs on Azure Government, and all of its customer data is processed in the continental United States. The processing map may shift as model availability changes.
Security Certifications and Trust Center
Independent attestation a buyer can pull without a sales call: SOC 2, ISO 27001, penetration test summaries, a trust center with current reports and named scope rather than a badge image.
An annual SOC 2 Type 2 audit covers security, availability and privacy, ISO 27001 certification covers the technical and operational delivery of the service, and penetration testing is commissioned. Data is encrypted in transit and at rest on AWS, and DISCO states GDPR compliance. The security page links a trust center at trust.csdisco.com, and the data processing addendum points to it for the subprocessor list. Customers can use two factor authentication, expiring passwords and single sign on through SAML, and a vulnerability reporting route is published. The auditing firm and the SOC 2 coverage period are not stated on the public pages.
ISO/IEC 27001:2022 and ISO/IEC 27018:2019 certifications, a SOC 2 Type II report, a SOC 3 report and the Cloud Security Alliance CAIQ are listed, with a request route for compliance certificates. aiR Government holds a FedRAMP Moderate authorization, the platform is assessed under IRAP at Protected, and Relativity states HIPAA compliance. Calder7, Relativity's in house security team, builds the Security Center monitoring application that customers use. Report dates and auditing firms are not published.
Model Supply Chain Disclosure
Which models sit underneath, whose they are, where they run, and whether the vendor commits to telling customers when that changes. A legal buyer inherits every dependency it cannot see.
Writer and OpenAI appear on the subprocessor list as large language model providers, Writer in the United States and OpenAI in the United States and the EU, with Pinecone as the vector database. AWS, Microsoft Azure and Google are named as hosting providers. The data processing addendum gives customers 14 days to object after the list is updated on DISCO's trust center. Model vendors must hold SOC 2 Type 2 or ISO 27001 certification and may not train on or retain customer data. No model or version is tied to a particular feature.
aiR runs on GPT-4o, GPT-4.1 mini, GPT-5 mini, GPT-5.1 and GPT-5.2, all through Microsoft Azure services. Which model or combination serves a task depends on the aiR system's purpose and on model capability and availability. Inference locations are published for every instance region, and customer data processed by aiR does not train or improve third party models. Relativity notes that those locations may change, and it promises no notice to customers before the models or locations change.
Commercial Transparency
Whether a buyer can learn what this costs without entering a sales process: published rates, the unit being charged, what sits behind an enterprise tier, and what implementation adds.
DISCO publishes its unit and structure without a figure. Its pricing page bills on processed data at a rate per GB, with no additional charge for unlimited use of Cecilia AI on the platform. The per GB rate includes filters, Cecilia Q&A, single document Q&A, document summaries, definitions, timelines and deposition management, and a feature table sets out what is included. The rate itself comes through a quote. The annual filing records the 2026 move to usage based pricing.
Pricing runs on a committed data volume, paid as you go or locked in on a one or three year commitment. aiR Review, Privilege, Case Strategy, Assist and claiR come with that volume, with no separate AI licenses and no per user fees. APIs, the App Hub and the MCP server carry no extra license cost, and specialized solutions are added as needed. Some aiR products consume aiR Units or carry fees set in the customer's contract. No rate per gigabyte or per unit is published.
Firm and Practice Coverage
Who the product is actually built for. AmLaw, midlaw, small firm and solo, in house departments, government and courts, and which practice areas are supported rather than merely claimed.
DISCO sells to law firms, corporate legal teams and government, with dedicated material for IP litigation, construction litigation, plaintiffs and a pro bono program. Its annual filing counts 1,549 customers, including 330 large customers, at 31 December 2025. The platform spans the ediscovery lifecycle with hold, ediscovery, review, timelines, deposition management and case building, and pricing is quoted for customers across the United States, the United Kingdom, Europe and parts of Asia. Firm size segmentation, language coverage and any statement of what the platform is not built for are not published.
Law firms, corporations, government agencies and legal service providers are the stated buyers, with 193 of the Am Law 200 among them and customers in more than 30 countries. Solution pages cover ediscovery, investigations, second requests, FOIA, data subject access requests, data breach response, early case intelligence and contracts, and plaintiff firms have a page of their own. aiR Government serves agencies under a FedRAMP Moderate authorization. What the platform is not built for, and which languages aiR handles, are not stated.
The 12 legal signals, side by side
Recorded rather than graded. These are the questions a practitioner has to answer before a tool touches a client matter, and the answers are taken from public material only.
Client Data in Training
Can material a lawyer puts into this product be used to train a model?
According to DISCO's AI page, the language models used in Cecilia never train on or learn from customer data and usage. The Auto Review page adds that its models are prohibited from training on or retaining customer data after processing. Both statements address the model layer, and the terms and conditions and AI supplementary terms contain no training clause. Nothing addresses whether DISCO itself trains or tunes any model on customer content.
Customer data processed by aiR systems is not used to train or improve third party models or services, which covers the GPT models aiR reaches through Microsoft Azure. Training of Relativity's own models is not addressed, and Prioritized Review learns from reviewers' coding decisions. No published agreement carries a training term.
Prompt and Output Retention
How long does the product keep what a lawyer typed, and can that be set to zero?
Auto Review's language models may not retain any customer data after processing, and every model vendor must hold SOC 2 Type 2 or ISO 27001 certification. On termination, the data processing addendum has DISCO delete or return customer personal data, subject to retention its own policies or the law require. No period is stated for how long DISCO keeps case documents, Cecilia prompts, generated answers or the explanation attached to each tag suggestion, and no customer setting for that window is described.
aiR Review writes a rationale and citations for every document, and aiR for Privilege drafts log descriptions, so each matter builds up AI generated commentary on its documents. No retention period, customer setting or deletion control is published for that output or for prompts.
Ethical Walls and Matter Segregation
Does retrieval respect the firm’s ethical walls, or can the model read across them?
DISCO describes no separation between users, teams or matters inside one customer and no ethical wall feature. Its published controls cover single sign on, two factor authentication and the vendor's own supplier and communications security. For a platform sold to large firms running concurrent matters, how access is screened between matters is not described.
Discrete permissions inside the platform govern what each user reaches, so a firm keeps matter access aligned in Relativity aiR rather than inheriting a document management system's rules. Lockbox blocks administrators, including Relativity staff, from customer data unless the customer grants access, and that access can be audited and revoked in real time. Ethical walls are not named as such.
Third Party Request and Subpoena Notice
If someone subpoenas the vendor for a firm’s data, does the firm hear about it first?
Under DISCO's data processing addendum, each party must notify the other promptly in writing of any subpoena or other judicial or administrative order seeking customer personal data, to the extent the law allows. No transparency report is published. The clause covers personal data, which in an ediscovery collection runs through most of the documents a firm uploads.
No customer agreement, data processing agreement or transparency report is published, so no commitment to notify a customer of a government or law enforcement request for its data can be read.
Primary Law Corpus Provenance
Where does the law in this product come from, and does the vendor have the right to use it?
No primary law collection sits behind the platform. Cecilia answers only from the customer's own evidence, and Auto Review classifies against tag descriptions written for each matter, so the source material is the documents collected in discovery. No vendor supplied legal content, license basis or update cadence applies.
aiR draws on the customer's own collected documents rather than a vendor supplied library of law, so no legal content source, license or update cycle applies.
Good Law Verification
Does the product tell you when the authority it just cited has been overruled?
Collected evidence, not published case law, is what DISCO works on, so its output cites documents, not authority. It describes no treatment signal and no check of later history.
aiR works on a matter's collected documents rather than published case law, and claiR answers from that same data with direct citations to it. No treatment signal or later history check applies to legal authority.
Refusal and Uncertainty Behavior
What does the product do when the answer is not in the corpus?
Nothing describes what Cecilia Q&A or Auto Review does when the evidence does not support an answer or a tag, or how uncertainty is shown to the reviewer. Cecilia answers only from the customer's evidence, with citations, and Auto Review explains each tagging decision.
aiR Review explains and cites each call, and aiR for Privilege ranks likely privileged documents by its prediction. No path is described for declining a call when a document does not support one, and no confidence threshold is shown to the user.
Fabricated Citation Record
Does a public court record exist addressing fabricated or hallucinated legal citations in output from this product?
Damien Charlotin's AI Hallucination Cases database, which tracks court decisions worldwide on hallucinated AI content, has no entry naming DISCO or Cecilia. DISCO answers and classifies from a matter's collected evidence rather than from case law.
The AI Hallucination Cases database maintained by Damien Charlotin, which records court decisions worldwide that address hallucinated AI content and the tool involved where known, has no entry naming Relativity or aiR. aiR works on a matter's own documents and cites the passages behind each call rather than citing case law.
Bar Guidance Alignment
Has the vendor engaged in public with the ethics opinions its buyers are bound by?
DISCO names no ethics opinion or bar guidance, including ABA Formal Opinion 512. Its Auto Review material speaks of defensibility, sampling and validation, but connects them to no rule of civil procedure or case law on technology assisted review.
No named ethics opinion, such as ABA Formal Opinion 512, or state bar guidance is engaged. The AI Principles commit Relativity to openness about its AI so that customers can defend their own processes.
Billing and Fee Posture
Does the vendor address what happens to the bill when the work takes an hour instead of six?
DISCO claims its savings in review hours, saying Auto Review completes a first pass review in hours with precision and recall above human reviewers, at up to 32,000 documents an hour. It bills on processed data per GB with unlimited Cecilia AI included, so the AI cost a firm passes on follows data volume rather than review hours. Nothing published addresses how AI assisted review is recorded or billed per matter, or whether clients are told.
In customer stories, JND reports cutting a corporate client's costs by nearly 60%, and Microsoft projects $1.6 million a year. The Redgrave LLP study recorded about 18 attorney hours against 1,123 for the same review. No per matter record of AI assisted work exists for billing, and nothing addresses how a firm bills or discloses that work to clients.
Outside Counsel Guideline Readiness
Can a firm get this vendor through a client’s AI clause without a bespoke negotiation?
A published subprocessor list names each provider with function and country, including Writer and OpenAI as language model providers, Pinecone as the vector database and AWS, Microsoft Azure and Google for hosting. DISCO's terms and conditions, AI supplementary terms and data processing addendum sit beside it. Certifications are stated with scope, SOC 2 Type 2 across security, availability and privacy and ISO 27001 for service delivery, and a trust center sits at trust.csdisco.com.
ISO 27001 and 27018 certifications, SOC 2 Type II and SOC 3 reports and the CAIQ are available through a request route, and the models and processing locations behind aiR are published. No subprocessor list, data processing agreement or client notice pack is published.
Court Disclosure Support
If a judge’s standing order requires an AI disclosure, can the product produce one?
Each Auto Review tagging decision carries an explanation on the document it concerns, and recall and precision are tracked in the product. The basis for a first pass review can therefore be reconstructed document by document. Cecilia answers carry citations to the evidence. No single export combines the model used, the sources retrieved and the human check, and no model is named for any feature.
Every aiR Review call explains why the document matters and cites the passages behind it, and aiR for Privilege drafts privilege log descriptions. Relativity names the models aiR uses and the processing location for each region. No single export per document brings together the model used, the passages relied on and the human review.
The questions both sides leave open
Derived from the records above rather than written, so it cannot favor either vendor. Take these into both conversations and ask each side the same question.
- Primary Law Corpus Provenance
- Good Law Verification
- Refusal and Uncertainty Behavior
- Bar Guidance Alignment
Which one fits
Choose DISCO if
- You want recall and precision tracked inside the review. Auto Review reports both as a team validates a first pass review, explains every tagging decision, and can be run by the customer or by DISCO's Managed Review team.
- You need the machine to show its work document by document. Auto Review gives a written explanation for every individual tag suggestion, Cecilia Q&A answers are crafted only from the customer's own evidence with citations always included, and agentic answers carry the logic, reasoning and citations each time, with Quick thinking and Advanced research modes letting the user choose how much autonomous reasoning runs.
- You want adoption figures with consequences attached. Because the operator is listed on Nasdaq, its numbers appear in a regulated annual filing rather than a web page: 1,549 customers including 330 large customers and a 98 percent dollar based net retention rate at 31 December 2025, alongside published tools for migrating data in from other discovery systems.
Choose Relativity if
- Your security review is the gate. Relativity's trust site names ISO/IEC 27001:2022, ISO/IEC 27018:2019, FedRAMP Moderate authorization, IRAP assessed at Protected, HIPAA, a SOC 2 Type II report, a public SOC 3 summary report and a Cloud Security Alliance questionnaire, with a published request route for compliance certificates and a named in house security team.
- You have to keep matters apart and keep the vendor out. Client domains isolate users, workspaces, groups and matters by client, and Customer Lockbox is on by default, restricting Relativity's own system administrators from reaching a customer workspace unless the customer explicitly grants access, alongside customer managed encryption keys and a Security Center monitoring application shipped to customers.
- Privilege review is where the risk sits. aiR for Privilege explains every decision, generates privilege log descriptions automatically and is positioned to guide counsel's second pass rather than replace it, and both come inside Relativity's committed data volume pricing rather than as separate licenses.
In summary
DISCO
DISCO is a cloud ediscovery and litigation platform spanning hold, ediscovery, review, timelines, deposition management and case building. Cecilia AI answers questions from case evidence with citations, and Auto Review performs first pass review at up to 32,000 documents an hour from tag descriptions written for each matter. DISCO says the models used in Cecilia never train on or learn from customer data. Its terms, data processing addendum and subprocessor list are published, and its annual filing reports 1,549 customers with 98 percent net dollar retention. Pricing is per GB of processed data with unlimited Cecilia AI included.
Relativity
Relativity operates its platform, now named Relativity aiR and formerly RelativityOne, on Microsoft Azure, covering legal hold and preservation, collection, processing, review, production and analytics. Its aiR suite of five generative products spans review, privilege, case strategy, data breach response and assistance. According to the AI Legal Index, Relativity's trust site names ISO/IEC 27001:2022, ISO/IEC 27018:2019, FedRAMP Moderate authorization, IRAP at Protected, HIPAA, SOC 2 Type II, a public SOC 3 report and a Cloud Security Alliance questionnaire. Lockbox keeps Relativity's own administrators out of customer data unless the customer allows it. Relativity says aiR data never trains third party models, and it publishes no retention terms or liability position.
Questions buyers ask
DISCO vs Relativity: which is better for ediscovery?
Relativity is stronger on its trust estate. It names a broad certification set, keeps its own administrators out of customer workspaces by default, and isolates matters by client through client domains. DISCO is stronger on measurement, publishing recall and precision for its first pass review, and on training, stating that its models are not permitted to train on customer data. From the AI Legal Index, based on each vendor's own published materials as of September 2, 2026. No vendor pays for placement.
Do either train their models on your case data?
DISCO says the language models used in Cecilia never train on or learn from customer data, and that Auto Review's models may not retain data after processing. Neither statement covers DISCO's own models, and its terms contain no training clause. Relativity says data processed by aiR does not train or improve third party models, and its own models are not addressed. From the AI Legal Index, based on each vendor's own published materials as of September 2, 2026. No vendor pays for placement.
What can you show a court about how the AI decided?
Both record a rationale for their AI decisions. DISCO produces a written explanation for every individual tag suggestion in Auto Review, so the basis for each classification across a whole first pass is recorded per document. Relativity attaches a rationale to every aiR decision and generates privilege log descriptions automatically, which is a court facing artifact produced by the AI work itself. Neither identifies which model produced a given output, and neither offers a single export combining model, sources and human verification. From the AI Legal Index, based on each vendor's own published materials as of September 2, 2026. No vendor pays for placement.
Which one publishes accuracy figures for AI review?
Relativity does. It publishes a Redgrave LLP study, with its method in a full report, in which aiR Review found 88 percent of responsive documents against 64 percent for active learning. DISCO says Auto Review delivers precision and recall 10 to 20 percent above human reviewers, with no test set or method, and tracks recall and precision inside the product for each customer's review. From the AI Legal Index, based on each vendor's own published materials as of September 2, 2026. No vendor pays for placement.
What do DISCO and Relativity both leave unpublished?
Neither publishes a customer agreement, so neither states an indemnity, a liability cap, a warranty on output or an insurance position, which matters most on privilege work where an inadvertent production is the consequential error. Neither names the models behind its AI. Neither states a retention period or deletion route for prompts, outputs or the explanations each system generates. Neither offers a data residency option. And neither engages the Federal Rules or any bar guidance on defensible AI assisted review. From the AI Legal Index, based on each vendor's own published materials as of September 2, 2026. No vendor pays for placement.
Neither vendor paid for inclusion, placement or a grade, and neither reviewed this page before it published. Everything on it comes from public material on the dates shown. How the index grades.
On training, Relativity says data processed by aiR does not train third party models and says nothing of its own models, and DISCO's statement likewise covers the model layer. DISCO's claim that Auto Review beats human reviewers by 10 to 20 percent carries no test set or method. DISCO publishes its terms, with an infringement indemnity, a 12 month cap and a separate data protection cap, while Relativity publishes no customer agreement. Neither vendor reviewed this page.