Irys vs StrongSuit: how they compare in 2026

Irys One researches case law, drafts motions, memos and contracts in a firm's voice, redlines against playbooks, analyzes documents across a matter through a knowledge graph built from the matter's files, and checks citations; Workbench agents gather material from connected email and files and return finished documents. Matter management, the document library and electronic signature support that work; without the models there is little left to sell.

The AI is the product. Every module StrongSuit sells is generative or agentic: research over the proprietary case database, memo and brief drafting, the assistant, contract drafting and redlining, discovery review, the timeline generator, proofreading and the oral argument simulator. Its home page describes the foundation as frontier models with reasoning grounded in retrieval. The case database is the one asset that would survive without the models, and it is sold only as the data the research agents retrieve from, not as a search product. The company was built around AI from its founding in 2023 as Callidus Legal AI.

Research runs against retrieved legal sources, and results link to the opinions and documents relied on. Cite Check checks identified citations against legal source databases, flags those it cannot match, and surfaces available treatment signals such as negative history. The Thorough mode checks case law citations against source as it writes, while the Quick mode is disclosed as not backed by citations. The Responsible AI page says plainly that no mode makes output correct and that Cite Check does not verify every citation. The company has published a run of its research system against a public legal agent benchmark of 1,251 tasks, with an 83.74% criteria pass rate and a 17.75% strict pass rate on all criteria. It also published its departures from the benchmark's method, and released the code and traces. That measures agentic legal work, not the accuracy of Irys One's citations.

Grounding is real and documented; the accuracy figures are asserted, not tested. Research output cites and links the authorities relied on. StrongSuit's home page claims 100 percent reference links to legal authorities and describes reasoning grounded in retrieval over a proprietary database of more than ten million federal and state cases with summaries and issue tags. Sample memos and a sample brief can be downloaded, and the Search Cases pages show the underlying case records. Case validation, added at the November 2025 rebrand, checks whether cited cases have been overturned. Against that, the home page publishes 0 percent hallucinations of case names in the research module with no test set, sample size, period or method. The FAQ refers to proprietary checks against hallucination and rigorous evaluations without publishing any. A Vals AI benchmark from February 2025 covered the product under the Callidus name; it is third party work that predates the current product.

The Quick mode is stated not to retrieve or verify authority and never to be the basis for authority cited in a filing; the Standard research mode is to be paired with Cite Check before its citations are relied on; the Thorough mode checks citations as it works. Workbench agents gather material and build documents, but emails are left in drafts and never sent without the user. Drafts arrive as tracked changes, Cite Check returns flags for an attorney to resolve, and an audit trail records what was run, when and in which mode.

StrongSuit commits in writing to having the models work alongside a supervising lawyer, with real review surfaces and no full control structure. Its home page section on keeping the lawyer in the loop says every draft, review and decision stays under the lawyer's strategic control. The FAQ describes work traditionally handled by junior associates while the user keeps final judgment, and the terms of service make the user solely responsible for reviewing, verifying and editing generated work before relying on it. The review surfaces are concrete: redlined Word files from proofreading, tracked edits through the Word add in, and inline citations on research output. No thresholds, modes or escalation rules are described, since the product does not act unattended, and nothing says what happens after an output is found wrong beyond the user's own correction.

The site quotes Omer Er, Partner at Michelman & Robinson, lists users at Falls Counsel, Cox O'Brien and the Law Offices of Li Yang, and says more than 300 legal teams use the platform. Its outcome figures, 41% fewer hours and turnaround cut from 3.9 to 2.3 days, are attributed to two unnamed legal teams in their first weeks on the product.

Named customers come without figures, and figures without named customers. Seven testimonials carry full names and roles: Erin Abrams, Chief Legal Officer at Via; Doug Johnston, a former general counsel; and five attorneys at solo or small firms, Yandy Reyes, Bryan Schwartz, Andrew Ayers, David Holt and Lindsey Lewis, each describing use in practice. Logos include Cooley and Klinedinst. None is tied to a measured change. The figures sit elsewhere, unattributed: 80 percent time savings on a first redline, a 30 page memo in 15 minutes, and a CEO quote in the November 2025 awards release that many attorneys report doubling their output. A claim of more than 1,000 legal organizations is a count, not evidence.

The Data Processing Addendum and Terms of Service commit that customer content is not used to train Irys's or third party models. It is kept separate for each tenant and user profile, never commingled with other customers, and deleted within 30 days of termination or earlier on request. The terms bind Irys to confidentiality, with prompt notice before any compelled disclosure. Third party models are named, used for stateless inference with retention on the provider side disabled, and bound by terms that bar training. Administrators cannot see a user's personal workspace unless it is shared. The security page presents the platform as designed to preserve confidentiality and support privilege, while the terms leave compliance with privilege obligations to the user. Separation by matter is described on product pages, not in the agreement.

Every part of the position is in writing before signing, and privilege is addressed in the agreement itself. On training, section 5.6 of the terms of service says StrongSuit will not train any AI model with User Content, and the privacy policy of 16 March 2026 extends this to third party models, including subprocessors. On privilege, section 5.3 acknowledges that user content may be privileged, deems StrongSuit a confidential agent of the user for support and maintenance, and says any access is not intended to waive privilege, invoking Federal Rule of Evidence 502. On separation, StrongSuit's FAQ says role based permissions confine access to specific projects or documents within an organization. On retention, unsaved inputs are discarded after output, and saved content is kept until the subscription ends and then deleted after a wind down given as, for example, 90 days. On providers, OpenAI, Google and Anthropic are named, processing occurs in a private tenant, and content is said never to enter their public training sets. Separation rests on an FAQ description rather than a documented mechanism.

The Terms of Service and a separate Disclaimer state that Irys gives no legal advice, creates no attorney client relationship and is an assistive tool for legal professionals. A Professional Responsibility and Ethics section requires independent judgment, review of material output including authorities, quotations and subsequent treatment, review before anything goes to a court, client or regulator, and firm policies, training and supervision for AI tools. A Responsible AI page explains in plain language why attorney review is not optional, and an AI Rules by State guide sets out disclosure requirements, ethics guidance, sanctions and CLE rules state by state, with sources. No consumer facing surface is published.

StrongSuit states plainly what the product is and is not, who may use it, and how it supports supervision duties. Section 2 of the terms of service states in capitals that StrongSuit is not a law firm, does not provide legal advice and does not practice law, and that the services work only as a research, drafting and review tool. They are intended solely for use by or under the supervision of a licensed attorney aged 18 or over, no attorney client relationship is created, and the user is solely responsible for reviewing and verifying generated work. The security and ethics page names ABA Formal Opinion 512, Model Rule 1.6 and Resolution 604, and says they require human oversight of AI generated work and reasonable safeguards. Jurisdiction is limited by scope: the case data is US federal and state law, and the terms bar use from sanctioned jurisdictions. There is no consumer product. The page also says the product is 100 percent compliant with every US state's ethical rules as of March 2025, which describes StrongSuit's own review rather than a published mapping.

The Responsible AI page sets out what Irys does and does not claim, how its safeguards are layered and what strong firm practice looks like; policies name Legal & Security at Iqidis, Inc. as their owner, and the security page says the architecture was audited by Guidepost Solutions. No accountable individual, internal governance framework, testing before release, or finding on whether output differs across matter types, jurisdictions or practice areas is published.

Statements on responsible AI come without a mechanism, a testing regime or anything open to audit. The security and ethics page has a section on Responsible Use of Generative AI, but its content is data handling: inputs are not trained on, not shared, and discarded after output. The home page refers to rigorous evaluations built for professional practice without publishing any. No governance framework, accountable owner, description of testing before release, or statement about uneven output across practice areas or jurisdictions is published, and no ISO 42001 or equivalent certification is claimed.

Canceled accounts stay exportable for up to 90 days and are then deleted within 30; the Data Processing Addendum commits to deletion within 30 days of termination or earlier on written request; users can delete data at any time. MFA is enforced for all accounts, with least privilege roles, SSO and SCIM, idle session expiry and an exportable audit trail. A subprocessor policy names every service provider and subprocessor with purpose, location and retention, and commits to 30 days' notice of changes with a right to object. Breaches are notified within 72 hours with set contents and equivalent obligations on subprocessors, and government requests are notified where lawful and challenged if overbroad.

Published policy covers most of the ground, short of an incident practice and a formal subprocessor list. On retention, the security page says inputs are processed to generate the output and then discarded unless the user saves them. Section 3 of the privacy policy keeps client data for an administrative wind down after the subscription ends, given as for example 90 days, then deletes it permanently, subject to legal hold; account data is deleted or anonymized within 90 days of closure, and billing records are kept seven years. Deletion is stated as unrecoverable. On access, the privacy policy describes two factor authentication for staff who touch customer data, role based permissions, a least access support policy, and audit trails on infrastructure and application activity. On subprocessors, Microsoft Azure for hosting and OpenAI, Google and Anthropic for models are named in prose in the privacy policy and section 6 of the terms, with an open ended "and others" and no list, locations or change notice. No breach notice commitment or incident process is published.

The Terms of Service cap each party's liability at twelve months' fees or $100, whichever is greater, but lift the cap for breaches of confidentiality and for gross negligence or willful misconduct. Irys defends customers against US patent, copyright and trademark claims over the service, including output used as delivered, and the Data Processing Addendum adds an indemnity for data subject or regulator claims caused by a security incident in Irys's systems. Irys does not warrant that output is accurate or that verification features catch every issue.

A published limitation clause disclaims the risk the product creates, so the allocation of loss can be read before signing. Section 7.1 of the terms of service disclaims all warranties, including any warranty about accuracy, completeness or reliability. Section 7.2 excludes indirect and consequential damages, including fines a court imposes, and caps total liability at fees paid in the twelve months before the claim. Section 2 has the user assume all risk for decisions taken on the output, acknowledge that output may contain errors, and hold StrongSuit harmless for them. Section 7.3 provides beta features without warranty or indemnity, and section 8 is an indemnity from the user to StrongSuit only. There is no vendor indemnity, insurance or warranty on output, and the terms say a signed master services agreement, where one exists, governs over them; none is published.

Irys runs inside Microsoft Word through an add-in on AppSource, and connects on the user's authorization to Google Drive, Gmail, OneDrive, Outlook and Dropbox for import and export, as listed in its subprocessor policy. Product pages also show iManage, NetDocuments, SharePoint and Slack, but its FAQ says document management integrations are in development, and what syncs and what a firm must configure is not documented.

The only integration is Microsoft Word. A Word add in is sold as the Negotiate module and on the pricing page as seamless Word integration, and the terms define software add ins, such as for Microsoft Word, as part of the services. StrongSuit's home page says it drafts contracts in Word and offers insights and clause suggestions on an open document. That describes a feature rather than documenting it for implementers. A Help Documents section exists. No document management, practice management, ebilling, filing or matter system integration is named, and the FAQ's answer on replacing a patchwork of tools describes consolidation inside StrongSuit rather than connection to the systems a firm already runs.

The subprocessor policy says data is processed primarily in the US and EU with regional routing on request, hosted and stored on AWS in the US, and gives each AI provider's processing location, US and EU for the main models. Product pages describe hard tenant isolation with containers for each tenant and each user, and name cloud, VPC and on premises options. What each deployment option involves, and which plans include which, is not described beyond Enterprise being priced to scope.

Tenancy and region are both stated, with partial residency detail. The security page says the platform is hosted in a private Microsoft Azure tenant with high availability across several data centers, and the privacy policy says all client data is stored on geographically redundant servers located only in the United States and that model processing occurs within a secure private tenant. That gives the tenancy model, a single region and the processing location. No region option beyond the United States is published, and no difference in deployment between the Individual and Firm plans, although the home page refers to flexible licensing and deployment without saying what varies.

The security page lists SOC 2 Type II, ISO 27001 (the subprocessor policy specifies ISO/IEC 27001:2022), HIPAA, GDPR and CPRA compliance, and an architecture audit by Guidepost Solutions; current reports and certificates are available to customers and prospects under NDA on request. A trust center is linked, and the security page and subprocessor policy publish no auditor, scope or report date.

StrongSuit states its certification, but the evidence is not accessible. The security page says the company is SOC 2 certified with independent third party verification, StrongSuit's pricing page lists SOC 2 Type II as a feature of every plan, and the home page carries a SOC 2 Type II image. No auditor, coverage period, scope or route to the report is published, there is no trust center or request portal, and the security page's offer to walk through architecture and protocols is a sales conversation. The control descriptions are specific: AES encryption at rest and in transit, two factor authentication, routine vulnerability assessments, blue green deployment and audit trails, plus an optional PII scrub and end to end encryption for customers with higher needs.

The subprocessor policy names OpenAI for inference and embeddings, Anthropic's Claude and Google's Gemini on Vertex AI for inference, Mistral AI for OCR, Perplexity AI and Linkup for research lookups and CourtListener for case law and citation lookup, each with processing location and retention, and commits to 30 days' notice of any new subprocessor or material change, with a right to object. It also describes the architecture: most requests are handled by Irys's own knowledge graph, retrieval and orchestration layers, with external models called only for discrete, stateless inference steps. Specific model versions are not listed.

Providers are named and the models are not. The privacy policy says the intelligence layer uses third party large language models from OpenAI, Google, Anthropic and others under contracts requiring confidentiality, with processing in a private tenant so data never enters the providers' public training sets, and that hosting is on Microsoft Azure. Section 6 of the terms repeats OpenAI, Anthropic and Google as outside providers whose terms the user must respect, and StrongSuit's home page says the platform is powered by frontier models. No model is named, the list is open ended, and there is no commitment to notify customers when a provider or model changes.

The pricing page publishes a free plan, Irys Lite and Irys Student at $19.99 a month, Irys One at $299 per seat a month or $3,049 a year, and Irys Teams at $329 per seat a month for 2 to 25 seats, with usage limits by plan and no metered overage on Irys One. Enterprise, for 25 or more seats, is priced to scope, and the enterprise deployment program is not priced.

Real prices are published for part of the range, and the firm tier is withheld. StrongSuit's pricing page lists an Individual plan at 249 dollars per month with its features, and a Firm plan on custom quote that adds onboarding, a success manager, centralized billing, user access management and collaboration. A five day free trial requires a card, and the FAQ repeats the figure and mentions volume discounts. Section 4 of the terms sets out the mechanics: monthly or annual automatic renewal, 30 days' notice of price changes, no refunds, a fair usage policy on plans marketed as unlimited, and usage caps on documents, module use and model tokens whose limits are not published. Nothing says what implementation adds, and the Firm price is not published.

Pages address solo practitioners, small, midsize and large firms, in house teams and insurance defense firms, with use case pages for litigation, corporate and contracts, insurance defense, employment, mass tort and ADA compliance, and law students have their own plan. Citation checking is described against federal and state opinions. The jurisdictions, courts and practice areas it does not support are not stated.

Coverage of segments and practice areas is described in substance, while the limits stay open. Solos and boutiques are served on the Individual plan and larger firms on the Firm plan, with in house use shown by a chief legal officer testimonial. Practice pages exist for family law, litigation and corporate work, with an All Practice Areas page and a Topical Expertise page, and the case data is US federal and state law, more than ten million cases. StrongSuit does not say where the product stops. No practice area, court or jurisdiction is named as unsupported, and coverage outside the United States is neither claimed nor excluded, beyond the terms barring use from sanctioned jurisdictions.