CaseMark vs Syllo: how they compare in 2026

Deposition summaries come in analysis, index, page and line, and narrative formats, and CaseMark says a 200 page transcript comes back in under two minutes. Medical chronologies list each provider by date with diagnoses, ICD codes, medications and vitals. A firm can send examples of its own chronology format and get it back as a one click template. Workspace adds chat across a matter's files, Deep Research, custom agents, persistent memory across conversations and matters, and Overtime Mode, which keeps researching and drafting after the first prompt. Tabular Analysis turns hundreds of documents into one table with a CSV or Excel export. In Source, models draft exhibit indexes, propose clip ranges from a plain language description of the testimony and answer questions across up to 50 transcripts. Insight Pass sits in front of every summary. It reads the selected sources, asks three or four questions about audience, side, purpose and detail, and hands a plan to the drafting models. Quick Actions run one click summarize, research, timeline, organize and find tasks, and Organize proposes a file structure and naming for a matter. Workspace usage is metered in money against a monthly allowance, and larger tasks consume more of it.

The capabilities the vendor leads with run on models, and the conventional layer underneath would still work without them. Agentic Issue Tagging coordinates several language models in distinct roles to run first and second level review with unlimited issue codes, and the open prompting engine answers questions across the case record and drafts summaries, deposition digests and chronologies. The vendor calls itself a Litigation AI platform, founded to build the data and workflow infrastructure into which language models are woven. Beneath that sits an eDiscovery module that stands on its own, with native and load file processing, Boolean and metadata search, bulk redaction and privilege log automation. Case management covers chronologies, exhibits, binders and transcript designations. Strip the models out and a litigation support team still has a review and case management platform.

Deposition summaries cite page and line and come as Word files or hyperlinked PDFs that open at the testimony. Every medical chronology entry cites its source page. Answers in Workspace's Explore view carry Bates citations, and Deep Research returns a synthesis with linked citations and a source list the user can review. After any summary, Interrogate lets a user ask follow up questions across the whole deposition or record set, with each answer linked to the supporting page. For research beyond the customer's files, case.dev's citation service matches each cited case against CourtListener's database of about 10 million opinions. It returns verified, not found or multiple matches, with the case name, court and decision date. No accuracy figure is published for any output. The terms warn that output may not reflect people, places or facts accurately and must be reviewed before anyone relies on it.

Accuracy is measured and published for document classification, and generated output is grounded but not measured. The white paper of 21 March 2025 reports that across the last ten completed responsiveness reviews in live litigation, estimated recall was lowest at 93.4 percent, averaged 97.8 and had a median of 99.4. Median estimated precision was 85.9 percent and the average 79.7, with figures per matter, dataset sizes, and the firms' own precision and elusion testing as the method. Every tagging determination carries a rationale for the issue that pinpoints the text behind it, and answers from the prompting engine carry hyperlinked citations to the case record. No figures are published for the accuracy of generated answers, summaries or digests. The failure modes the paper names belong to competing approaches, such as prompt overload, context windows and hallucination in linear review, not to this system. The benchmark is eighteen months old at this date. Grounding is to the customer's own record, not primary law, so checking citation status does not arise.

In Source, the exhibit index starts as a draft that the court reporter edits, matches to matter files and approves. Clip ranges proposed from a plain language request come back with the matched testimony, a confidence value and warnings, and a clip is cut only once approved. The reporter keeps sole authority over the certified transcript, and Source prepares what surrounds it. Summaries can be aimed before drafting through Insight Pass, and every output links back to its sources. CaseMark describes a chronology as an organized starting point whose dates, statements and conclusions still need confirming against the underlying records. At the platform layer, case.dev tracks each unit of agent work as a work item that a person approves or blocks with a stated reason. Its privilege detection service returns a review recommendation, calling for attorney review, on ambiguous documents. Two modes run with no person in the loop. Overtime Mode keeps researching, drafting and revising after the first prompt, and the Operator handles recurring tasks it has learned through email, Microsoft Teams and text messages. Neither has a stated limit on what it may do unattended.

Review surfaces and a validation route are real, and no threshold or use constraint is published. The white paper describes the review system as coordinating several language models that decide how to conduct the review within guidelines users set. The product page says case teams validate the outcomes of agentic issue coding to ensure defensibility for outgoing productions, and tag conflict detection is built into quality control. The white paper's case studies show that validation as precision testing on tagged sets and elusion testing on the null set. Customer and Case Administrator roles control who can act on a case. Nothing says at what point the system stops or escalates, what any mode's output may be used for, or what happens after a coding error reaches a production. The home page headline claims the platform can autonomously manage any matter end to end; that describes the Litigation Agent, in early access, and no oversight structure is published for it.

Ogletree Deakins, Nelson Mullins and Pathway are named as users of CaseMark's platform, and CaseMark counts more than 7,000 attorneys among its users. Fortz Legal Support, a Michigan court reporting and litigation support firm, ran one deposition through several summary vendors alongside peer agencies before choosing CaseMark. After one summary for a client, that client's partner asked for summaries of every remaining deposition in the case, and Fortz now sells summaries as a service. A September 2026 post reproduces one month's usage report from an unnamed mid sized insurance defense firm, for 12 August to 11 September. It shows 38 practitioners working 98 litigation matters, 155 deliverables, 1.25 million pages loaded and 841 attorney hours saved, for a flat $3,500 a month. CaseMark values those hours at about $106,000 on the firm's blended rates and calls that an assumed value. A medical malpractice insurer is described as saving $5 for each $1 spent across 300 panel firms in 36 states. The solutions pages quote results such as 90 percent less document review time for small firms and $11,500 in average savings per case for insurers, without a basis. In the Chock Barhoum LLP story, two weeks of document review shrank to minutes. The partner it features, Sarah Tuthill-Kveton, is also CaseMark's Chief Legal Officer.

Named firms, a date, figures for what changed and a method that can be assessed are all published. The 21 March 2025 white paper lists practitioner contributors from seven firms, among them Ballard Spahr, Royer Cooper Cohen Braunfeld and Quinn Emanuel, and reports results per matter measured by the firms' own teams. Ballard Spahr's trial team tested a review of more than 100,000 documents against more than 25 issue codes and found estimated precision of 95.56 percent and recall of 99.4 percent, after about three hours of setup. Royer Cooper Cohen Braunfeld compared approaches on a random sample of just under 16,000 documents from more than 150,000. Its managed review team using continuous active learning reached estimated recall below 67 percent at over two dollars a document, against the system's recall of 93.44 percent and precision of 69.81 percent. The paper is published by the vendor, the matters are not named, and the contributors wrote in their individual capacities. Its footnotes disclose that one technical contributor is a research advisor and one practitioner a product advisor to the vendor.

Training on customer data is ruled out in the terms, and the security page says documents, transcripts and work product remain the customer's. Under the terms, the customer keeps ownership of uploaded data and any interest in the output. Inside a firm, each matter is shared with read or write permission, and a Document Manager seat covers file work without chat or billing access. Custom agents are scoped to the organization, a matter or one person. CaseMark describes the platform as tenant isolated with no commingling of customer data, and Enterprise adds dedicated compute and isolated inference pools. Model providers sign BAAs with CaseMark, and the case.dev gateway serves Anthropic's models from a zero data retention workspace. CaseMark says the architecture itself protects privilege, through tenant isolation, encryption and audit trails. No published policy covers privilege or work product in what the models see or produce.

Privilege and work product are treated expressly in writing, and training is left open. The Legal Disclosure and Subpoena Policy, incorporated by the terms at section 9, states that much of what users store is confidential and privileged under the attorney client privilege and the work product doctrine. The terms add that Content is not shared with third parties except service providers, legal process and security monitoring. Vendor personnel may not view Content without permission or a support, security or legal reason under internal controls (sections 13.1 and 16.6.2). Segregation is documented by case. The license in section 8.4.1 covers improving the service, and no commitment against model training is published. The position of any outside model provider is not stated.

A No Legal Advice clause in the terms says CaseMark does not practice law, that text from its AI tools is for information only, and that no attorney client relationship forms. Users must be 18 or older and may not present output as written by a person. The service is offered only to people in the United States, and personal data covered by the GDPR may not be uploaded without CaseMark's consent. The products are sold to lawyers, paralegals and litigation support staff rather than to the public. Medical chronologies are described as built for legal team review and not as a substitute for medical or legal judgment. No published material sets out how the product supports a lawyer's supervision of the work it produces.

Professional responsibility is placed on the user and the audience is widened, while the marketing speaks of legal analysis and strategy. Section 5.4 of the terms says the service is not intended for use solely by attorneys, judges, paralegals or others in the legal profession and may be used by anyone holding a licensing agreement. It adds that the vendor has no responsibility to determine or monitor whether a user is engaged in the unlicensed practice of law. Section 5.8 makes each user responsible for being authorized to practice where they use the service for the practice of law. The product is marketed for work product generation, legal analysis and case strategy, and the home page headline speaks of autonomously managing a matter. No statement says the product does not give legal advice, nothing addresses a supervising lawyer's competence or supervision duties, and no jurisdictional limit is stated beyond the US account rule.

A September 2026 post by CEO Scott Kveton sets out what CaseMark weighs when it chooses a model provider. The factors are retention, the reach of a BAA, training terms a provider can change, and dependence on one vendor. In July 2026 CaseMark signed the Open Weights and American AI Leadership letter. Its CEO argued that a model a vendor hosts itself cannot be withdrawn, repriced or quietly retuned under a customer. The agent layer is open to inspection. Linc, CaseMark's terminal agent, is MIT licensed and built on the open source pi agent framework. Its legal skills follow an open standard published through agentskills.legal, and custom skills a firm commissions are portable to other platforms. No results from testing its summaries or models are published.

A measurement practice is published, and no governance framework, owner or testing before release surrounds it. The white paper describes validating outputs in live matters and reports aggregated results across ten reviews, and it names a Carnegie Mellon Language Technologies Institute researcher as a technical contributor. The Work with Syllo page says the AI systems are designed for rigorous human review and feedback. The terms separate Beta Previews from the audited service, stating that early access features are not subject to the same security measures and auditing. No responsible AI policy or principles page is published, and no person or function is named as accountable for model behavior. Nothing describes what is tested before a model or feature is released, and uneven output across matter types or parties is not addressed.

Customer data is kept for 30 days or for the length of the agreement, then destroyed through a process CaseMark says follows ISO standards. Users can delete a file and its chat themselves, and deletion on request is also offered. Access follows least privilege, with user roles, SAML single sign on and multifactor authentication required on every account. Every user action and system event is logged. Data is encrypted with 256 bit AES at rest and TLS 1.2 in transit. External penetration tests run every year, vulnerability assessments every quarter and dependency updates daily. Staff pass background checks and annual security training. CaseMark publishes more than 130 security controls with their status, among them disaster recovery plans tested each year and cybersecurity insurance. A public status page reports uptime and outages, and payments run through Stripe. No commitment on breach notice to customers is published.

The Work with Syllo security section states encryption in transit and at rest in line with NIST SP 800-57, automatic backups, continuous vulnerability scanning and penetration testing, and multifactor authentication. The terms require credentials for one person only and user notice of a compromise within one business day (section 2.1), and aim for 30 days' notice before a feature is discontinued so users can export (16.3). Outside the product policies, no commitment to notify customers of a security incident is published, and no list names who processes content for the AI features. The terms place a product Privacy Policy and Data Security Policy at app.syllohq.com/policies, a page that opens only after signing in to the product.

The terms of service, last updated 1 October 2026, cap CaseMark's total liability at $100, and more than one claim does not raise the cap. Lost profits, lost data and indirect damages are excluded. The service is provided as is, with warranties of accuracy and noninfringement disclaimed, and any warranty the law requires lasts 90 days. CaseMark's suppliers carry no liability under the same clause. The indemnity runs from the customer to CaseMark, covering use of the service, uploaded data and breaches of law or of the terms. No indemnity runs the other way, and wrong output carries no remedy. Enterprise customers sign a separate agreement on top of these terms, and that agreement is not published.

Liability is handled through a disclaimer, a nominal cap and an indemnity running to the vendor. Section 15 provides the service as is and as available, disclaims any warranty that content provided through it is accurate, reliable or correct, and excludes all categories of damages. It caps the vendor's total liability to an end user at one hundred dollars and excludes liability for breaches at third party providers. The indemnity runs the other way, from each user to the vendor, for their use, their content and their breaches. Nothing addresses AI output specifically, no vendor indemnity is offered, and no insurance position is published. The terms bind end users and give way to the customer's Licensing Agreement where the two conflict, so negotiated terms may differ, but that agreement is not published.

Inside Workspace, files arrive by upload and leave by download or by saving back to the matter. CaseMark's own comparison with Harvey lists practice management connectors as coming soon in the product. The case.dev platform underneath publishes connectors in depth. Clio Manage is generally available. It imports matter documents, communications and tasks into a vault and exports work product to a CaseMark Output folder in Clio. Two way sync keeps documents from each side separate and halts on conflicting edits instead of overwriting either. Google Drive needs enablement and Box is in preview, and change detection, rate budgets and reconnection are set out for each provider. Developers reach the platform through an API with SDKs for TypeScript, Python and Go, a command line tool, signed webhooks and an MCP server for Claude, ChatGPT and Microsoft 365 Copilot. Source exports the litigation support formats that trial software reads, including MDB, PTX, LEF and SBF.

The product stands alone by design. On its own platform page, the vendor argues that API integration and acquisition have both failed to fix fragmented litigation tooling, and offers a single environment holding the whole record instead. What it supports is interchange: native file and load file ingestion with deduplication, threading and family grouping, load file generation and Bates stamping at production, and import of standard court reporting formats in text, audio and video. Those formats make moving a matter onto the platform routine, but no connection to an external system is named. No document management system, Microsoft Word or Outlook plugin, matter management, electronic billing or court filing integration is published, and no API or developer documentation is published.

Customer data sits in AWS and Azure in the United States. The trust center adds that it can be stored in other countries when a customer needs that. CaseMark hosts its open weight models in the United States. Isolation rises by tier. The standard service isolates each tenant, and Enterprise adds dedicated compute and isolated inference pools. Sovereign Legal AI on case.dev offers CaseMark Core on a dedicated GPU cluster, dedicated instances of frontier models, or a managed deployment inside the customer's own AWS, Azure or Google Cloud account. The processing location for closed models from OpenAI, Anthropic and Google is not stated. When OpenAI is down, the gateway routes requests to Azure or Bedrock.

Residency is offered and the hosting provider is named, and the deployment options are not specified. The security section says multiple secure deployment options are available, data storage is regionalized to customer needs, and for firm engagements the vendor will work with a firm's security team to customize deployment. The terms name the infrastructure: Content is stored on Amazon Web Services virtual servers and Amazon S3 storage the vendor controls, with ElasticSearch for indexing (section 13.3). User accounts may not be used outside the United States, a restriction the terms tie to privacy and data protection law (sections 2.1 and 5.6). The options themselves are not named, so whether dedicated, private cloud or on premises arrangements exist cannot be established. No regions are listed, and nothing separates where model inference runs from where data is stored.

A SOC 2 Type II report and HIPAA compliance are stated. The trust center, built on Vanta, lists the 2026 SOC 2 Type 2 report and a statement of HIPAA compliance as documents released through a request form. The same center shows more than 130 controls across infrastructure, organizational, product and internal security, each with its status. The report itself, with its auditor and period, sits behind that request form.

A SOC 2 Type II attestation is named with its scope and auditor, and no route to the report exists. The compliance section states SOC 2 Type II certification in Security, Confidentiality, Availability and Privacy, compliance with the AICPA SOC standards for service organizations, and third party validation by Prescient Assurance. No trust center is published on the site or by search on this date, no report period or issue date is given, and no request route of any kind is described.

The case.dev gateway offers CaseMark's own Core Large, Core Titanium, Core Mini and Embed V1 models. It also offers OpenAI's GPT-6.1 Sol, GPT-6 Sol and GPT-6 Luna, Anthropic's Claude Opus 5.5, Sonnet 5.5 and Haiku 5.5, and Google's Gemini 3.1 Pro and Gemini 3.8 Flash. Workspace lets a user pick from more than 100 models, with a cost multiplier shown for each. By its CEO's account, about two thirds of CaseMark's inference runs on open weight models, and its embedding service runs a self hosted open weight model. Anthropic's models are off by default in Workspace and are switched on for firms that ask. Insight Pass uses a reasoning model to plan each report and lighter models to draft it. No commitment covers telling customers when a model or provider changes.

The architecture is described in detail, and no provider or model is named. Product pages say the platform runs on a combination of proprietary Syllo models and foundational language models. The white paper describes an ensemble of language models of varying size in distinct roles, such as strategizing, quality control and resolving disagreement between models, with telemetry routing harder material to more capable models and simpler material to cheaper ones. No provider, model family or version is named on any surface read, nothing says where inference runs as distinct from the storage named in the terms, and nothing addresses what a model provider may retain. On change, the terms reserve the right to modify any feature at any time with or without notice (section 13.7).

Summaries are sold one at a time with no subscription, at $25 for a legal summary. A medical chronology costs $25 plus $0.01 a page, capped at $100 for one record set of up to 25,000 pages. Transcription costs $5 an hour with a $5 minimum. Workspace starts with a free trial for one user with $5 of usage and no card. Pro is $100 per user a month with $100 of included usage, billed monthly with no annual term. Work beyond the allowance draws on prepaid on demand usage at a 30 percent premium, capped by a team spend limit. Document Manager seats are $20 a month, and a BAA plan starts at $200 per user a month with two custom workflows per firm. Teams, for two to ten users, and Enterprise are quoted. So is Source, which is priced on the certified packages, workflow runs and clips a firm produces. The case.dev developer platform passes model costs through at provider rates and opens with $10 of free usage.

No price, unit of charge, tier or list of inclusions is published. The pricing section of the Work with Syllo page names two customer segments and routes both to a sales conversation: corporate legal teams are offered the platform on an enterprise basis, and law firms a firmwide enterprise partnership. Firmwide describes the scope of a license, not a unit anyone could price. The home page, platform and tagging pages, the terms and the white paper were also searched, and the only cost figures on the site describe competitors.

Workspace is sold to solo and small firms, mid sized firms and Am Law scale teams, with in house teams named among its intended buyers. Source serves court reporting and litigation support firms. The solutions pages cover litigation, personal injury, corporate, regulatory and compliance, real estate and family law, plus insurance carriers on the defense side. The workflow library follows that spread, with 663 litigation, 279 transactional, 182 regulatory, 139 corporate and one family law workflow. Record types include depositions, medical records, pleadings, contracts, recordings, and trial, hearing, arbitration and mediation transcripts. Service is limited to the United States, and the Clio connector supports only Clio's United States region. CaseMark's own comparison names bulk review at Vault scale and the breadth of research sources as areas where Harvey is stronger. CaseMark runs monthly Workspace training, with a recorded session aimed at insurance defense. Government buyers are not addressed.

Segments and matter types are described with evidence, and the boundaries are not. The vendor names AmLaw 100 firms, litigation boutiques and corporate legal teams managing complex litigation portfolios, with a dedicated page for the last. The white paper shows the breadth with named firms from AmLaw defense practices to a plaintiffs' employment firm. It lists thirteen kinds of matter the system has been used on, among them antitrust, patent and mass tort, across datasets from thousands to more than two million documents. The scope is litigation throughout. Nothing addresses government or court use or names matter or data types it handles poorly, and the one stated limit, the US account rule, is framed as data protection rather than coverage.