ContractSafe vs ContractWorks: how they compare in 2026
ContractSafe and ContractWorks are both contract repositories with AI tagging and unlimited users, sold as simple contract management to organizations that want contracts stored, searchable and tracked, and both add drafting and approvals on higher tiers. ContractSafe publishes much more about its terms of sale: prices from $450 a month by contract volume, no implementation charge and a 5 percent annual cap on increases, while ContractWorks lists document limits and signature licenses and leaves its plan prices blank. ContractSafe also states how its AI behaves, with a person accepting each extracted value and review that never edits the contract, and says customer contracts do not train AI. ContractWorks runs under Onit's agreement and subprocessor list, which name OpenAI as the AI provider, let Onit use aggregated, anonymized customer data for product improvement, warrant that the service performs as documented, and state that Onit is not a law firm. Both hold SOC 2 Type II reports, and ContractSafe's is downloadable after an instant NDA.
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.
ContractSafe is a contract repository first: storage, tagging, custom fields, unlimited reminders and alerts, dashboards and reports, with approvals, redlining, Word editing, eSignature, templates and intake forms on higher plans. AI sits across it as a set of features. AI data extraction suggests key terms and dates for a person to accept, correct or skip; AI Search takes plain English queries; Ask AI answers questions about an open contract; and AI Contract Review checks incoming contracts against a playbook of the customer's rules. ContractSafe's FAQ says customers can use some of the AI features or none at all, and the repository, search by tag and field, alerts and workflow all work without them. AI extraction and contract chat come with every plan, AI extraction for custom fields and AI Contract Review come with Maximize, and ContractSafe also sells Jump Start, where its paralegals extract data from existing contracts by hand.
ContractWorks is a secure contract repository with search, tagging, alerts, reports and eSignature, and on Premium, drafting, redlining and approval workflows. AI is a layer on that base. Smart Document Tagging adds tags such as effective date, renewal date and termination date on upload, Smart Contract Generation identifies agreement type, parties, effective dates, renewal terms and notice windows, search is described as AI powered, and Premium's feature list includes a GPT-3 AI Clause Creator. The repository, folders, custom tags, OCR search, highlights, date alerts, reporting and signatures all work without the models, and ContractWorks also sells a contract tagging service performed by its legal engineers and licensed attorneys. The product pages describe it as a secure, AI powered repository for teams that need simple, effective contract management.
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.
ContractSafe's AI answers from the customer's own contract and points back to it. Ask AI gives answers sourced from the open contract, AI Contract Review links each failed rule to the relevant clause with a plain explanation of why it failed, and extracted values appear as suggestions beside the document for a person to accept or correct. Review runs against rule definitions the customer writes, with preferred and fallback language, rather than a generic template. The AI page says the AI isn't perfect and that results are easy to review and adjust. No accuracy figure, error rate, test set or evaluation is published for extraction, search, answers or review, and no benchmark is cited. Release testing is described as internal testing followed by a beta with customers who opted in and a staged rollout.
ContractWorks publishes no accuracy figure, error rate or evaluation for its AI tagging, extraction, search or clause creation. Snippet View lets a reviewer see tagged information in context across several agreements at once, which lets a person check a tag against the contract text, and drafting is marketed as producing compliant, error free contracts without any measure behind it. Section 12.3 of Onit's Subscription and Services Agreement, last revised 19 August 2026, says AI output is machine generated and may be incomplete, inaccurate, biased or unsuited to the customer's purpose, and makes the customer solely responsible for review and reliance. The GPT-3 clause creator names a model generation that OpenAI has since replaced, and nothing says whether the feature now runs on a newer model.
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.
ContractSafe sets out what each AI mode may do. Extraction only suggests values: a person accepts, corrects or skips each one before it is added, singly or in bulk across contracts. AI Contract Review reads and evaluates the contract and does not edit it; suggested wording is copied into a redline only if the team chooses. Rules that AI drafts for a playbook must be reviewed and accepted by a reviewer before they are added. AI search returns only contracts the searcher is allowed to see. The AI page states the principle as the AI suggests and a person decides, customers can switch AI features off entirely, and feedback on any result can be given at any time. Every user action is recorded in a timestamped audit trail. Nothing published describes a confidence threshold or what happens to a value that was accepted in error.
Section 12.3 of Onit's services agreement makes the customer solely responsible for review of and reliance on AI output, including obtaining legal or other professional advice as appropriate, and says Onit is not liable for decisions taken on it. In the product, AI tags are added automatically on upload and can be reviewed in Snippet View, and Premium approval workflows route contracts under automated conditional flows. Administrators alone invite users and set each person's permissions, and an audit trail records every click. Nothing published says whether AI tags are applied without review, whether a person confirms an AI drafted clause before it enters a contract, or what happens when a tag or alert date is wrong. Section 12.8 says no service levels apply to AI features unless an order says so.
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.
The City of Olympia case study names the customer, a state capital with 600 employees serving 50,000 residents, and quotes Sean Krier. Olympia reports 75 percent less time spent scanning documents, about $10,000 a year saved in staff time, and a rise from 1 to more than 4 on a five point internal survey of contract management. It chose ContractSafe after reviewing more than 20 systems and trialing two. The rollout coincided with the move to remote work at the start of the pandemic, the only date given. Other named voices include Margaret Howell Benson, General Counsel, Pieter-Francois Theron, CFO of JourneyApps, Donna Yslas of World Oil and Neka Rodriguez of Radiologic Associates of Fredericksburg, with qualitative quotes. ContractSafe says more than 1,900 organizations use it, and the AI features carry no customer outcome figures.
ContractWorks lists named case studies for Esquire Deposition Solutions, BNP Paribas, Walmart Canada, DNA Payments, Slack, Sandoz, Carlsberg, OLX Group, Louis Dreyfus Company and GreenShield Canada. Named testimonials quote Louis Belzil, General Counsel of Tier 1 Energy Solutions, Lynn Towler, Senior Contracts Manager at Shift4 Payments, Nicole Dunn, Contracts Administrator at the Chesapeake Bay Foundation, and Adrienne Foster, Contract Manager at Xeris Pharmaceuticals, who reports loading more than 1,200 contracts in the year to date. The workflow page claims 80 percent faster contract workflows, at least 100 hours saved, at least $9,000 in value and a two week rollout, with no customer or method behind the figures. The case study titles give no deployment dates, and the outcomes quoted concern ease of use, price and repository setup rather than the AI features.
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 Terms of Use, effective 22 January 2025, treat customer content as Confidential Information protected with the care ContractSafe gives its own, and leave ownership with the customer (2.3, 3.1). Clause 3.2 licenses ContractSafe to use content to provide and improve the services, extends that license to trusted third parties as needed, and states that those third parties will not use the content to train their AI models. The AI page says ContractSafe does not use customer data to train AI models, and the security page says confidential contract data is never used to train public AI models. Inside an account, permissions run down to individual folders, tags, users and teams. Privilege and work product are not addressed, and a data processing addendum is sent on request rather than published.
Onit's Subscription and Services Agreement, last revised 19 August 2026, leaves the customer owning its data (5.1) and treats it as confidential, with reasonable advance notice before a compelled disclosure where law permits (8.5). Section 5.2 lets Onit use customer data, once aggregated and anonymized, for product improvement, analysis and benchmarking, section 12.1 lets it use deidentified usage data to improve its AI, and the same section bars third party model providers from training on input for others' benefit except in deidentified or aggregated form needed to run the service. Nothing states whether Onit trains its own models on customer contracts. The ContractWorks security page says ContractWorks never has backend access to customer data, and adds watermarks and view only access. Privilege and work product are not addressed.
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.
ContractSafe's Terms of Use, privacy policy, AI pages and security page contain no statement on the line between the product and legal advice, and no reference to lawyers' professional duties. The product is sold to education, healthcare, nonprofits, municipal government and small businesses as well as legal teams, and its users include executive assistants and compliance coordinators. Its AI answers questions such as whether a contract can be assigned and suggests replacement wording when a contract fails a playbook rule. The nearest statements are about oversight rather than advice: the AI suggests and a person decides, and review is applied against rules the customer writes.
Section 4.14 of Onit's services agreement, headed No Legal Advice, states that Onit is not a law firm and does not offer legal advice, recommendations, opinions, representation, referrals or counseling, and takes no responsibility for the quality or compliance of documents the customer generates. Section 12.3 tells customers to obtain legal or other professional advice as appropriate when relying on AI output, and section 12.7 bars using the AI features for decisions with a material legal or similarly significant effect on an individual. ContractWorks is sold to HR, sales, finance and nonprofit teams as well as legal, and its contract tagging service is run by legal engineers and licensed attorneys. Nothing addresses how the product supports a lawyer's competence and supervision duties or any limit by jurisdiction.
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.
ContractSafe describes how an AI feature reaches customers: internal testing, a beta with customers who opted in, a check against its security and efficiency requirements, and release in small batches, with customers able to give feedback on results afterward. Its subprocessor list names two AI monitoring services, Braintrust and Langfuse, alongside the model providers. On regulation, the AI page says ContractSafe takes reasonable steps to review changes in laws affecting privacy and AI, including the EU AI Act, without describing what it has done. No person or team is named as accountable for AI, no test results are published, and nothing addresses whether output differs across contract types, languages or regions.
Neither ContractWorks nor Onit publishes an AI governance position: there is no responsible AI page, named owner for model behavior, release testing process or result, and no AI management certification. The AI terms in Onit's services agreement allocate risk rather than describe governance. Section 12.3 warns that output may be biased, section 12.4 lets Onit improve, replace or deprecate models or prompts without materially reducing core functionality, and section 12.7 restricts the customer's own uses. The SOC 2 Type 2 report covers security, availability and confidentiality of systems rather than model behavior, and nothing addresses whether tagging performs differently across contract types or languages.
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.
ContractSafe's security page describes hosting on Amazon Web Services with AES-256 encryption at rest and in transit, a web application firewall, continuous offsite database backups, a disaster recovery plan, continuous intrusion detection, daily vulnerability scans, regular penetration testing, Sophos endpoint protection and a timestamped audit trail of user actions. Access controls include SAML single sign on, two factor authentication, role based permissions to the document level and optional IP allowlisting. A subprocessor list names each AI provider and service with its regions. Clause 9.4 of the Terms makes content available for 30 days after termination, after which ContractSafe may delete it, and the privacy policy keeps data while the account is active. No breach notification commitment or incident response practice is published, and the data processing addendum is sent on request.
The ContractWorks security page describes TLS 1.2 with 256 bit encryption in transit, 256 bit AES at rest, salted and Bcrypt hashed passwords, hourly encrypted backups across data centers in the United States and the European Union, SAML single sign on, multifactor authentication by SMS or authenticator app, administrator controlled permissions, watermarks, view only access, and an audit trail of every click that can be exported or emailed daily. A security specification PDF is downloadable. Onit's subprocessor list, updated 17 August 2026, names the suppliers for the SecureDocs platform that runs ContractWorks, and customers can subscribe to change notices with 15 days to object under section 4.5 of the agreement. The pricing FAQ says records are deleted within 30 days after a customer leaves, and section 6.5 points to a retention policy that is not published. No breach notice period is stated.
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.
The Terms of Use provide the services as is, with no warranty of accuracy (11.1), and commit to commercially reasonable efforts toward 99.9 percent monthly availability (2.1). ContractSafe defends and indemnifies the customer against third party claims that the services infringe a valid US copyright or US patent, excluding the customer's content and modifications, with replacement, modification or a prorated refund as remedies (11.4). Each party's liability is capped at fees paid under the subscription in the 12 months before the claim, with indemnities and intentional misconduct outside the cap (11.3). The customer indemnifies ContractSafe for its use of the services and its content (11.5). Nothing addresses loss from a wrong AI extraction, answer or suggestion, and no insurance is published. Disputes go to binding arbitration under California law.
Onit's Subscription and Services Agreement, last revised 19 August 2026, warrants that the subscription services operate substantially as documented and that material functionality will not decrease during the term (9.2). Onit defends and indemnifies the customer against claims that the services infringe intellectual property rights (11.1). Each party's liability is capped at fees paid or payable in the 12 months before the event (10.2), with listed claims, including indemnities, outside the cap (10.3). For AI, section 12.5 provides the AI features and output as is, with all warranties disclaimed, section 12.3 says Onit is not liable for decisions taken on output, and section 12.8 excludes AI features from service levels unless an order says otherwise. No insurance is published.
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.
ContractSafe's plan table lists a DocuSign integration, single sign on, Zapier and a Microsoft Word integration on Finalize, and APIs and a Salesforce integration on Maximize. The Word integration is described in direction: edit in Word with tracked changes and sync the result back to ContractSafe. Contracts can also be sent in by email, which the privacy policy says runs through Cloudmailin, and Dropbox Sign appears as an eSignature utility on the subprocessor list. ContractSafe's own team sets up integrations for customers rather than leaving the work to them. The integrations page lists no further detail, and no API reference, field mapping for Salesforce or document management connection such as iManage or NetDocuments is published.
ContractWorks lists integrations as features in its plan table: a Salesforce integration, Fujitsu ScanSnap scanning, uploading contracts by email, SAML 2.0 single sign on, and on Premium, integration with public registers and databases. The security page names Microsoft Active Directory, OneLogin and Okta as identity providers for single sign on. Premium drafting works in a Word compatible editor, and the AXDRAFT platform that runs Premium lists DocuSign as its signing API. No integration page describes what the Salesforce connection syncs, in which direction or what an administrator configures, no API is documented, and no document management system used by law firms is listed. Onit's subprocessor list names Workato as an integration platform across Onit products.
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.
ContractSafe runs on Amazon Web Services and offers data residency in the United States, Canada, the European Union and Australia. Its pricing FAQ says that once a region is chosen the data stays entirely within it, and that an account can sit in only one region. AI processing is listed by region on the subprocessor page: Microsoft Azure OpenAI, AWS Bedrock and OpenAI in the US, EU, Canada and Australia, Google Gemini in the US only, and the Langfuse monitoring service in the US and EU, with Braintrust in the US. Email intake through Cloudmailin is processed in the United States and the European Union. Nothing says which AI feature uses Gemini, and no single tenant or on premises option is described.
Onit's subprocessor list, updated 17 August 2026, says ContractWorks runs on the SecureDocs platform on Amazon Web Services, with the customer choosing hosting in the United States or Ireland and backups in Germany. ContractWorks Premium runs on the AXDRAFT platform, where AWS hosting can be in the European Union, Canada or the United States and Google Cloud in the EU is the primary infrastructure. OpenAI is listed for all Onit platforms, with its locations given as the United States and worldwide. The security page says data is stored in multiple data centers in the United States and the European Union. No tenancy model, private deployment or on premises option is described.
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.
ContractSafe lists SOC 2 Type II certification with an annual audit, ISO 27001 certification, TX-RAMP Level 2 certification, HIPAA compliance and GDPR compliance on its security page. Its trust center, hosted on Vanta, shows real time security posture and lets a prospect sign an NDA instantly and download the full SOC 2 Type II report without a sales call. The pricing FAQ says SOC 2 is certified by an independent audit firm, without naming it. Clause 8.2(j) of the Terms of Use requires anyone who receives a SOC or penetration test report to keep it confidential and use it only to evaluate ContractSafe. Vulnerabilities can be reported to a published security address. The ISO 27001 version, certificate scope and auditor are not shown on the public pages.
ContractWorks states a SOC 2 Type 2 report from an independent auditor covering security, availability and confidentiality, and makes a security specification PDF available for download. The ISO 27001 certification and SOC 1, 2 and 3 reports named on the security page belong to the AWS data centers it uses, not to ContractWorks itself. No auditor, report period or route to request the SOC 2 report before signing is published, and no trust center is linked. Onit's data processing addendum lets customers rely on recent SSAE 18 reports in place of an on site audit.
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.
ContractSafe's AI page says it uses models from OpenAI, Google and Amazon. Its subprocessor page lists the AI providers with their regions: Microsoft Azure OpenAI, AWS Bedrock and OpenAI in the US, EU, Canada and Australia, and Google Gemini in the US, plus Braintrust and Langfuse for AI monitoring. Clause 3.2 of the Terms of Use says third parties that receive content will not use it to train their AI models. No model name or version is given, nothing maps a provider to a feature, and no commitment to notify customers before a provider changes is published.
Onit's subprocessor list names OpenAI, in the United States and worldwide, as the AI provider used across all Onit platforms, and ContractWorks Premium's feature list names a GPT-3 AI Clause Creator. Onit's contract products listed separately use Anthropic and Google Gemini, but neither is listed for ContractWorks. Customers can subscribe to email notice of subprocessor changes and object within 15 days under section 4.5 of the services agreement, while section 12.4 lets Onit replace models or prompts without notice if core functionality is not materially reduced. Nothing says which ContractWorks features use OpenAI or which model version runs the tagging and search.
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.
ContractSafe publishes prices. Organize starts at $450 a month, Finalize at $660 and Maximize at $815, prepaid annually, for up to 100 active contracts, and the price rises through published volume bands of 101 to 500, 501 to 1,000, 1,001 to 2,500, 2,501 to 5,000, 5,001 to 10,000 and above 10,000. Every plan includes unlimited users, onboarding and training, data and document migration, unlimited archive and attachments and a dedicated customer success manager. Implementation costs nothing. Only contracts in the repository count, not archived drafts or attachments. A monthly option is offered by card. Jump Start data entry by ContractSafe's paralegals starts at $2.50 per contract. Clause 1.4 of the Terms caps price increases at 5 percent a year for the same plan.
ContractWorks publishes three plans with their limits but leaves the price fields blank. Standard includes unlimited users, 2,500 documents and five electronic signature licenses; Professional includes 10,000 documents and ten licenses; Premium adds drafting, redlining and workflows with unlimited documents. All are billed annually. Additional signature licenses cost $1,200 a year for a pack of five, a nonprofit discount is offered, and a 14 day trial converts to the Standard license. The pricing FAQ says template automation and onboarding are part of a paid implementation, while the workflow page says ContractWorks handles the rollout at no extra cost. The FAQ also uses older plan names, Essential, Standard and Enterprise, for the signature allowances. Orders are signed electronically and paid by card, ACH or wire.
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.
ContractSafe has pages for education, healthcare, hospitality, nonprofits, small businesses, IT and software, and teams with dispersed staff and vendors, and its named customers include a city government, an oil company, a radiology practice and a software company. It describes itself as built for companies of all sizes and says more than 1,900 organizations use it. Legal is one of several buyers; quotes come from a general counsel, a CFO, an executive assistant and a compliance coordinator. TX-RAMP certification and a HIPAA compliant offering support Texas public bodies and healthcare. Nothing published sets a size limit, names contract types or languages the AI handles poorly, or addresses law firm use.
ContractWorks has solution pages for HR, sales, healthcare, SaaS, construction and engineering, government, nonprofit and hospitality teams, and describes itself as built for teams that need simple, effective contract management, with unlimited users on every plan. Its case studies span retail, banking, payments, software, pharmaceuticals, consumer goods, commodities and legal services, and testimonials come from a general counsel, contracts managers and a contracts administrator. Premium is positioned for companies scaling to a full CLM. Clients outside the United States subscribe and pay in US dollars. Nothing published names contract types or languages the AI tagging does not handle, sets a size limit or addresses law firm use.
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?
The FAQ on ContractSafe's AI page answers whether customer contracts train AI models with a plain no, and the security page says confidential contract data is never used to train public AI models. The Terms of Use, effective 22 January 2025, go part of the way: clause 3.2 licenses ContractSafe to use customer content to provide and improve the services and extends that license to trusted third parties, which will not use the content to train their AI models.
The agreement does not restrict ContractSafe's own training in the same words. The privacy policy, effective 15 December 2023, says ContractSafe may use analytic means to evaluate, train and improve services that process content.
Onit's Subscription and Services Agreement, last revised 19 August 2026, which governs ContractWorks, bars third party model providers from using input to train their models for the benefit of others, except in deidentified or aggregated form necessary to operate the service (12.1). The same section lets Onit use deidentified, aggregated usage data to operate and improve its AI services, and section 5.2 lets it use customer data, once aggregated and anonymized, for product improvement, analysis and benchmarking.
No opt out is described, and nothing states whether Onit trains its own models on customer contracts. The ContractWorks pages say nothing about training.
Prompt and Output Retention
How long does the product keep what a lawyer typed, and can that be set to zero?
The privacy policy keeps information while the account is active, as needed to provide the services and to meet legal obligations, and says some data may be deidentified and kept. Clause 9.4 of the Terms of Use makes a copy of the customer's content available on request for 30 days after termination, with metadata as a CSV file and documents in their native format, after which ContractSafe may delete it. Authorized users can download all documents and data at any time.
No retention period is stated for AI questions, answers, extracted suggestions or review results, and nothing states what the AI providers keep.
Section 6.5 of Onit's services agreement gives the customer 30 days after termination to export its data and then deletes it under Onit's current retention policy, which is not published. The ContractWorks pricing FAQ says all records are deleted within 30 days once a departing customer confirms it is off the system, and customers can bulk download their documents at any time. The security page describes hourly encrypted backups across data centers in the United States and the European Union.
Section 12.2 says Onit has no duty to retain or recreate AI output unless the service expressly provides retention and it is enabled. No period is stated for AI tags, clause suggestions or search queries during the term, or for what OpenAI keeps.
Ethical Walls and Matter Segregation
Does retrieval respect the firm’s ethical walls, or can the model read across them?
ContractSafe keeps its own permission model and publishes its parts in the plan table: custom roles and permissions, access permissions to folders and tags, permission groups, user level permissions and team level permissions, on every plan. The security page says access can be controlled down to the individual document, and adds SAML single sign on, two factor authentication and optional IP allowlisting. The AI page states that AI search returns only the contracts the searcher is allowed to see, so the AI respects the same permissions at query time.
Read only users and full administrators are both counted as users with no extra charge. Nothing is published on separation between customers beyond hosting on AWS.
ContractWorks keeps its own permission model. Only administrators can invite users to the repository, and they choose permission settings for each person, with access controlled by folder and document, permission based user roles in every plan, view only access that disables printing, and personally identifiable watermarks. Sign in can run through SAML single sign on with Microsoft Active Directory, OneLogin or Okta, with multifactor authentication by SMS or authenticator app.
An audit trail records every click with user, date, time and action and can be emailed to administrators daily. The security page says ContractWorks never has backend access to customer data. Nothing published says how AI search and tagging apply folder permissions when they run, or how customers are separated from one another on the platform.
Third Party Request and Subpoena Notice
If someone subpoenas the vendor for a firm’s data, does the firm hear about it first?
ContractSafe's privacy policy states a policy of challenging government access to data in court before disclosing it, says it will attempt to notify data subjects of any such request where the law allows, and commits to commercially reasonable efforts to notify users about law enforcement or court ordered requests unless prohibited. It adds that nothing in the policy limits a user's own defenses or objections to a request.
Clause 2.2 of the Terms of Use binds ContractSafe to adhere to its privacy policies, and clause 2.3 lets content lose its confidential status where disclosure is required by law, subject to the privacy policy. The policy also reserves release of personal information under legal compulsion, such as a subpoena. No transparency report is published.
Section 8.5 of Onit's Subscription and Services Agreement, last revised 19 August 2026, which covers ContractWorks customer data as confidential information, commits to reasonable advance notice of a disclosure required by law, where law permits, and reasonable cooperation, at the customer's expense, in seeking confidential treatment, and says the disclosed information stays confidential between the parties afterward.
Onit's data processing addendum commits to notify the customer promptly of government requests about its processing unless prohibited. The ContractWorks security page adds that ContractWorks never has backend access to customer data. No transparency report is published.
Primary Law Corpus Provenance
Where does the law in this product come from, and does the vendor have the right to use it?
ContractSafe's AI works from the customer's own contracts and rules. Ask AI answers from the open contract, AI Contract Review evaluates a contract against rule definitions the customer writes, with its preferred and fallback language, and extraction reads the uploaded document, with each PDF converted to text first. Playbooks can be started from a negotiated contract, highlighted language, a sample playbook or one suggested by contract type, and AI drafted rules are accepted by a reviewer before use.
The product does not retrieve case law or legislation. The models come from OpenAI, Google and Amazon, and nothing describes their training data or the basis on which any data behind them was used.
ContractWorks' AI reads the contracts a customer uploads to tag dates, parties and terms, builds an outline from section headings after OCR, and supports search across the repository. It does not retrieve case law or legislation. Nothing published identifies what the tagging models were trained on or what sources the GPT-3 clause creator draws on, and Onit's subprocessor list names OpenAI as the AI provider without describing its training data.
For customers who want tags applied by people, ContractWorks sells a contract tagging service run by its legal engineers and licensed attorneys.
Good Law Verification
Does the product tell you when the authority it just cited has been overruled?
ContractSafe manages a customer's own contracts and does not cite case law or legislation, so a citator is not part of the product. Nothing on the AI, review, security or pricing pages addresses checking authority for later history. Playbook rules and fallback language are written and kept current by the customer, and nothing describes how the product flags a rule that a change in law has made out of date.
ContractWorks stores and tags a customer's own contracts and does not cite case law or legislation, so a citator is not part of the product. Nothing on the ContractWorks pages or in Onit's agreement addresses checking authority for later history, and the Premium clause library is maintained by the customer with nothing describing how outdated language is flagged.
Refusal and Uncertainty Behavior
What does the product do when the answer is not in the corpus?
ContractSafe describes review rather than abstention. Extracted values are suggestions a person accepts, corrects or skips, AI Contract Review shows pass or fail per rule with an explanation, and suggested wording is offered when available. No confidence score is shown, and nothing says what Ask AI does when the answer is not in the contract or what extraction does with a field it cannot find. The AI page says the AI is not perfect and that results are easy to review and adjust.
No path for declining to answer is documented, and no confidence score is shown for AI tags, search results or clause suggestions. Section 12.3 of Onit's services agreement says output may be incomplete, inaccurate or biased and leaves review to the customer, and section 12.4 lets Onit change models or prompts without notice. Snippet View shows tagged values in context across agreements so a person can check them. Nothing says what tagging does with a date it cannot find or a clause it cannot classify, which matters because tagged renewal and termination dates drive the date alerts.
Fabricated Citation Record
Does a public court record exist addressing fabricated or hallucinated legal citations in output from this product?
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 ContractSafe. This is a statement about the public record rather than a finding about the product, and it covers fabricated content only. ContractSafe manages contracts rather than producing court filings, so its output does not ordinarily reach a brief.
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 ContractWorks, SecureDocs or Onit. This is a statement about the public record rather than a finding about the product, and it covers fabricated content only. ContractWorks stores and tags contracts rather than producing court filings, so its output does not ordinarily reach a brief.
Bar Guidance Alignment
Has the vendor engaged in public with the ethics opinions its buyers are bound by?
No bar opinion, ethics rule or reference to lawyers' professional obligations appears in ContractSafe's Terms of Use, privacy policy, AI pages, security page or pricing FAQ, and there is no statement that the product does not give legal advice. The AI page's regulatory statement concerns privacy and AI laws such as the EU AI Act, which bind ContractSafe as a provider rather than a lawyer using the product.
Section 4.14 of Onit's services agreement states that Onit is not a law firm and does not offer legal advice, recommendations, opinions, representation, referrals or counseling, and section 12.3 tells customers to obtain legal or other professional advice as appropriate when relying on AI output. No bar opinion, ethics rule or professional conduct guidance is named on the ContractWorks pages or in Onit's agreements, and nothing connects the product to a lawyer's duties when using AI tags or clauses.
Billing and Fee Posture
Does the vendor address what happens to the bill when the work takes an hour instead of six?
ContractSafe is bought by organizations to manage their own contracts, and its named customers are a city, companies and a medical practice rather than law firms, so no client bill sits in the loop. Its savings claims are aimed at the buyer's own time and cost, such as the City of Olympia's 75 percent less time scanning documents and about $10,000 a year saved in staff time, and its pricing pitch rests on unlimited users and no implementation charge.
Jump Start, where ContractSafe's paralegals extract data from a customer's existing contracts from $2.50 per contract, is a service sold to the customer, not a fee passed to anyone else. Nothing published addresses fee treatment of AI assisted work for a firm that bills a client.
ContractWorks is bought by organizations to manage their own contracts across legal, HR, sales, finance and nonprofit teams, and its named customers are companies and a nonprofit rather than law firms, so no client bill sits in the loop. Its savings claims are aimed at the buyer's own time, such as 80 percent faster contract workflows and at least 100 hours saved. Its contract tagging service, run by legal engineers and licensed attorneys, is a service sold to the customer. Nothing published addresses fee treatment of AI assisted work for a firm that bills a client.
Outside Counsel Guideline Readiness
Can a firm get this vendor through a client’s AI clause without a bespoke negotiation?
ContractSafe publishes a subprocessor page that names its AI providers with regions, Microsoft Azure OpenAI, AWS Bedrock, OpenAI and Google Gemini, its AI monitoring services, Braintrust and Langfuse, and HubSpot and Dropbox Sign. The privacy policy names Cloudmailin, Pendo and Invoiced with their processing locations. A data transfer impact assessment and a data protection policy are published. The data processing addendum is sent on request, and the SOC 2 Type II report is downloadable after an instant NDA in the Vanta trust center; clause 8.2(j) of the Terms of Use requires anyone who receives a SOC or penetration test report to keep it confidential and use it only to evaluate ContractSafe. No client facing AI disclosure material is published.
Onit publishes a dated subprocessor list, updated 17 August 2026, with a table for the SecureDocs platform that runs ContractWorks and a table for the AXDRAFT platform that runs ContractWorks Premium, and names OpenAI as the AI provider across all platforms with its locations. Customers can subscribe to email change notices. The services agreement and data processing addendum are published, and the addendum lets customers rely on recent SSAE 18 reports in place of an on site audit.
ContractWorks offers a downloadable security specification PDF. A customer can forward these to its own clients without an agreement in place.
Court Disclosure Support
If a judge’s standing order requires an AI disclosure, can the product produce one?
ContractSafe records a complete, timestamped audit trail of user actions across all contracts, presented as support for internal compliance reviews and external audits, and edits made in the app are versioned. Because a person must accept each AI extracted value and choose whether to use suggested wording, the record of who added a value exists. Nothing published says whether the audit trail marks a value or a clause as AI suggested, or which model produced it, and no export or template for disclosing AI use is published. The output is contract data and reports rather than court filings.
ContractWorks records every click in an audit trail with user, date, time and action, which administrators can export to Excel or have emailed daily, and Premium keeps version control across redlines with third parties invited to collaborate without signing in. Nothing published says whether a tag or clause added by AI is marked as machine generated in that trail, or which model produced it, and section 12.2 of Onit's agreement says Onit has no duty to retain AI output unless the service provides retention and it is enabled.
No export or template for disclosing AI use is published. ContractWorks output is a contract repository and reports rather than court filings.
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.
- Good Law Verification
- Refusal and Uncertainty Behavior
Which one fits
Choose ContractSafe if
- You want the price before the demo. ContractSafe publishes Organize from $450, Finalize from $660 and Maximize from $815 a month, prepaid annually, priced by contract volume with unlimited users and no implementation charge. Its Terms of Use cap increases at 5 percent a year for the same plan.
- You want the AI to suggest and a person to decide. ContractSafe has a person accept, correct or skip each extracted value, its contract review never edits the document, rules drafted by AI need a reviewer, and the AI features can be switched off. AI search returns only contracts the searcher is allowed to see.
- Your data has to stay in Canada or Australia. ContractSafe offers residency in the US, Canada, the EU and Australia with each account held in one region, and lists its AI providers by region. Its SOC 2 Type II report is downloadable from its trust center after an instant NDA.
Choose ContractWorks if
- Your contracts need watermarks and view only access. ContractWorks adds personally identifiable watermarks, view only access that disables printing, and an audit trail of every click that administrators can have emailed daily. Only administrators invite users and set each person's permissions.
- You want the subprocessor list with a change notice. Onit's subprocessor list, updated 17 August 2026, covers the platform that runs ContractWorks, and customers can subscribe to change notices and object within 15 days. Onit also publishes its services agreement and data processing addendum.
- You want a performance warranty in the agreement. Onit's services agreement warrants that the service operates substantially as documented and keeps its material functionality during the term, and indemnifies the customer against intellectual property claims. Liability is capped at 12 months of fees, with indemnities outside the cap.
In summary
ContractSafe
ContractSafe is contract management software built around a searchable repository and renewal tracking, adding approvals, Word editing, eSignature, templates and intake forms on higher plans. Its AI extracts key terms for a person to accept, answers plain English questions about a contract and reviews contracts against the customer's own playbook rules. According to the AI Legal Index, ContractSafe publishes in detail on price and control: plans start at $450, $660 and $815 a month with unlimited users, each AI mode's limits are stated, and data can stay in the US, Canada, the EU or Australia. Its terms carry no statement that it does not give legal advice, and no breach notice commitment is published.
ContractWorks
ContractWorks is a contract repository sold by Onit for teams that need simple contract management. AI tags each uploaded contract with details such as agreement type, parties, effective date, renewal terms and notice windows, and the product adds AI search, OCR, alerts, scheduled reports and built in eSignature, with drafting, redlining and approvals on Premium. According to the AI Legal Index, ContractWorks runs under Onit's published agreement, which states that Onit is not a law firm, warrants performance, indemnifies against infringement and lets Onit use aggregated, anonymized customer data for product improvement. Its plans publish document limits and signature licenses but not prices, and no AI governance material is published.
Questions buyers ask
ContractSafe vs ContractWorks: which is better for a small team?
Both suit a small team that mainly needs a searchable repository with alerts and unlimited users, and both hold SOC 2 Type II reports and sell human data entry for existing contracts. ContractSafe publishes its prices and its AI controls and offers more hosting regions. ContractWorks adds watermarks and view only controls and runs under Onit's agreement with a performance warranty. From the AI Legal Index, based on each vendor's own published materials as of October 8, 2026. No vendor pays for placement.
How much do ContractSafe and ContractWorks cost?
ContractSafe starts at $450, $660 and $815 a month, prepaid annually, rising with contract volume, and its Jump Start data entry starts at $2.50 per contract. ContractWorks publishes plan limits but no plan prices; its one published figure is $1,200 a year for each extra pack of five signature licenses, and it offers a 14 day trial and a nonprofit discount. From the AI Legal Index, based on each vendor's own published materials as of October 8, 2026. No vendor pays for placement.
Do ContractSafe or ContractWorks train AI on customer contracts?
ContractSafe says it does not use customer data to train AI models, although its Terms of Use bar third party training only. Onit's agreement for ContractWorks lets Onit use aggregated, anonymized customer data for product improvement and bars third party model providers from training on input except in deidentified or aggregated form. Nothing published states whether Onit trains its own models on customer contracts. From the AI Legal Index, based on each vendor's own published materials as of October 8, 2026. No vendor pays for placement.
Where are ContractSafe and ContractWorks hosted?
ContractSafe runs on AWS with residency in the US, Canada, the EU or Australia, and lists Microsoft Azure OpenAI, AWS Bedrock and OpenAI in all four regions and Google Gemini in the US. ContractWorks runs on AWS in the US or Ireland, with backups in Germany, and Onit lists OpenAI for AI processing in the US and worldwide. ContractWorks Premium runs on Onit's AXDRAFT platform, with Google Cloud in the EU as its primary infrastructure. From the AI Legal Index, based on each vendor's own published materials as of October 8, 2026. No vendor pays for placement.
What do ContractSafe and ContractWorks both leave unpublished?
Neither publishes an accuracy measure for AI tagging or extraction, a model version, a named owner for AI behavior, or anything on legal privilege and work product. Neither says whether its audit trail marks a value as AI suggested. ContractSafe's data processing addendum is sent on request, and the Onit retention policy that governs deletion after termination is not published. From the AI Legal Index, based on each vendor's own published materials as of October 8, 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.
ContractWorks is sold by Onit and governed by Onit's Subscription and Services Agreement, last revised 19 August 2026; its Premium tier runs on Onit's AXDRAFT platform. ContractSafe's training statement sits on its AI and security pages, while its Terms of Use bar third party training only. Both sell human data entry for existing contracts: ContractSafe's Jump Start from $2.50 per contract and ContractWorks' contract tagging service run by legal engineers and licensed attorneys. Neither vendor reviewed this page.