Juro vs SpotDraft: how they compare in 2026
Juro and SpotDraft are two midmarket contract lifecycle platforms sold to the same in house team, and they differ by disclosure rather than by function. Juro publishes the paperwork and the model layer. Its master services agreements for United States and non United States customers, standalone AI terms carrying a version history, a data processing agreement and an EU Data Act addendum are all dated and downloadable. It also states plainly that its AI runs on OpenAI's GPT model provided by Microsoft on Azure servers. SpotDraft answers on where the data lives and on professional guidance. It states logical separation within a multi tenant infrastructure, with personal data held in a customer selected region across the United States, the European Union, India and the Middle East and not transmitted outside it. Its servers run on Google Cloud in the Netherlands, with per contract encryption keys. It also names two ethics guidance documents.
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.
Juro manages the complete contract lifecycle, from template creation and approval routing through negotiation, electronic signature, storage and tracking. AI Review redlines third party contracts, AI Extract captures data from signed agreements, and the AI Assistant supports drafting and summaries. Operator adds conversational access to the repository and can answer questions with source citations. The underlying contract platform also has a distinct function without those AI features. Legal teams control templates and approval processes while colleagues generate agreements within those controls. Native electronic signature and lifecycle tracking complete transactions independently of an AI response. The company began in 2016 and added AI to that established contracting workflow.
SpotDraft is a full contract lifecycle system, with templates, conditional workflows and approvals, a shared editor, built in eSignature meeting ESIGN, eIDAS and ECA, a repository, reporting and analytics. Without SpotDraft AI, VerifAI, Intake and Sidebar, a working CLM with signature and workflow remains, and that CLM has its own market. The AI covers review inside Word, automatic extraction of more than a thousand metadata types, and agents that track regulatory change. SpotDraft now calls itself context aware, AI native CLM, though the platform predates that framing.
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.
Generated output may misstate people, facts or laws, including legal authorities, Juro's AI terms warn. Operator connects repository answers to source citations, while AI playbooks constrain tasks with contract context and customer rules. Prompt shortcuts help users structure instructions consistently. These features provide context and a route back to source material, but no measured hallucination rate, legal accuracy benchmark or test set accompanies them. Juro does not explain the retrieval mechanism behind Operator's citations. Juro treats generative AI as fallible and places responsibility for identifying and correcting errors with the customer before reliance. A source citation therefore provides a review aid, without a separate guarantee that the answer is correct.
VerifAI connects each playbook guideline to source references in the contract and explains why a requirement is met or unmet. It then proposes wording, comments and tracked edits against the customer's preferred positions. That provides a contract level basis for inspecting a recommendation. SpotDraft's broader AI offering also promises explained recommendations with citations, while Sidebar monitors official regulatory sources. These claims do not include a reproducible accuracy benchmark or a described citator for case law. The AI agreement explicitly anticipates inaccurate or incomplete internet material and requires independent validation, so source references are a review aid rather than a guarantee of legal correctness.
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.
Human review is required before customers use or rely on AI output. Juro also prohibits presenting generated output as human authored. Inside the product, legal teams establish approval routes and set playbook context and constraints at the template or individual contract level. AI Review explains which playbook rule supports a proposed edit, and reviewers can accept, edit or skip suggestions. The responsible AI guidance limits recommended use to tasks that the professional could perform and verify independently. These controls put the reviewer between a proposed redline and its acceptance. Juro does not say what stops unattended agent activity or how an erroneous result is escalated.
Users constrain VerifAI through their own playbooks or supplied guides, inspect the explanation for each deviation and receive edits as tracked changes in Word. Public guides let business colleagues conduct an initial review before legal involvement. The consolidated AI agreement requires human oversight and makes the customer responsible for checking outputs before business or legal decisions; it also excludes specified high risk uses. Conditional CLM approvals and permission roles provide an additional organizational control. VerifAI's nutrition label nevertheless marks human involvement as absent. The binding verification requirement and editable review surface are concrete controls, while the label is an unresolved inconsistency in product messaging.
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.
Customer outcomes come attached to named legal and business leaders. Talentful reports an 86 percent reduction in time to sign, Funnel an 88 percent reduction in manual contract reviews, and Luno a 91 percent time saving per contract. Paddle attributes three and a half hours saved per contract to the platform. Goldin's stated workload is 20,000 contracts annually, while Deliveroo describes eight years of use. These examples cover contract operations across multiple companies and roles. None isolates the contribution of an individual AI feature, and the figures lack a measurement method or deployment period. Older lifecycle results cannot establish the performance of the current AI review and Operator features.
Eight in house lawyers are quoted by name, role and employer. Anna Claveria Brannan is Deputy General Counsel at IPSY. Susan Koenig is a former Senior Legal Operations Manager at Abnormal Security, and Micah Nessan a former General Counsel at Guideline. Reason Abajuo is VP of Legal and Corporate Affairs at Chaberton Energy, and Lizzy Gagan is Senior Legal Counsel at Beamery. Arzu Hasanova is Legal Counsel at Circularise, Aditi Kapoor is Director of Legal at Gameskraft, and Natasha Wilson is Head of Legal at SUN Mobility. Every quote is qualitative. The quantified claims name no customer. They are two times faster closings, 65 percent lower cost, 70 percent less review time and contracts reviewed 15 times faster.
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.
Customer data cannot train an AI model serving another customer. Juro's AI terms permit a model used exclusively for the same customer, with deletion of that model within 60 days after termination. Generated output belongs to the customer and is treated as customer data. The responsible AI guidance identifies OpenAI's GPT family through Microsoft Azure and describes confidentiality obligations covering Juro and its technology providers. A data processing agreement and named subprocessors support that contractual framework. Repository separation and permission enforcement across legal, sales, HR and procurement are not explained in that material. Professional confidentiality is recognized, but attorney client privilege and work product receive no separate treatment.
SpotDraft's consolidated agreement treats uploaded customer material as confidential and expressly forbids using customer data as a training dataset for machine learning or model training. Its DPA restricts personal data processing to customer instructions, protects generated personal data and provides isolation between Sidebar customer environments. Contract encryption keys and roles further limit access. The AI terms also state that using the service creates no attorney client relationship or related privilege beyond contractual confidentiality. Generated material may be retained to debug reported issues, and provider specific retention is not fully explained. These commitments improve confidentiality assurance without establishing complete treatment of every privileged prompt and output.
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.
A person must review generated output under Juro's rules, which prohibit passing it off as human authored. Its responsible AI guidance recognizes lawyers' professional confidentiality duties and extends a related concern to HR teams handling employee information. The contracting workflow allows sales, procurement, HR and finance teams to act within legal team templates and approvals. These are operational controls around contract production, rather than a jurisdiction specific professional responsibility framework. The terms and guidance contain no statement that the product does not give legal advice, identify no governing bar opinion and do not map product behavior to ABA Formal Opinion 512. Nor do they restrict which jurisdictions or categories of professional work the AI can support.
The consolidated AI terms identify SpotDraft as a software provider, disclaim a lawyer relationship and make customers responsible for independent verification before legal or business decisions. Users must be told that AI is involved and cannot present generated output as human authored. Sidebar additionally prohibits processing sensitive financial or personal data and information about children under thirteen. These are concrete use boundaries alongside the company's references to California bar guidance and ABA Formal Opinion 512. They do not map each product control to a professional duty or establish a jurisdiction by jurisdiction competence framework for legal work performed through the platform.
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.
Privacy, confidentiality and accuracy in contracting are the focus of Juro's responsible AI framework. Customers can obtain a data protection impact assessment template populated for AI Assistant, with separate data protection guides for AI Assistant and AI Extract. AI playbooks let legal teams specify context and constraints, while an EU AI Act guide addresses the regulatory setting. The framework acknowledges bias and intellectual property issues but treats them as less pertinent than its three principal concerns. It does not identify a person accountable for model behavior, a release testing regime or evaluation results. Those omissions leave the operational ownership of model risk and the testing of biased outputs unclear.
A formal risk governance policy approved by management defines an Enterprise Risk Management program, according to SpotDraft's security page. Periodic operational risk assessments feed management reports, with each risk rated, assigned an owner and tracked to treatment or acceptance. Privacy risk is assessed through vendor due diligence, and an information security team led by the Chief Technology Officer oversees the process. The policy does not cover model behavior. SpotDraft's own web pages describe no testing before an AI release and no responsible AI framework. They do not address bias or uneven output across contract types, counterparties or populations.
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.
Juro names infrastructure providers including Algolia, AWS and MongoDB, along with analytics, communications and payments providers. Salesforce, Slack and Google also appear as customer initiated integrations. A data processing agreement governs personal data, and customers have access, portability and erasure rights. Separate security and resilience guidance accompanies a status page and a 99.9 percent uptime commitment. The AI terms require deletion of customer specific models within 60 days of termination; that period is not a retention schedule for contract text, prompts or generated output. Juro gives no broader content retention window and no AI specific incident notification process.
The DPA requires notification within 48 hours of awareness of a personal data breach, advance notice of subprocessor changes and an objection process. It provides return of customer personal data within thirty days and deletion within ninety days of return or as directed, with legal retention exceptions and a deletion certificate. Access reviews, encryption and annual testing are specified. The subprocessor register now identifies the AI providers as well as hosting and support services. These contractual controls cover substantial ground, but the AI terms permit retention of generated material for debugging without a complete separate schedule for nonpersonal prompts, outputs and provider logs.
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.
Juro separates its AI terms from the Master Services Agreements for customers in the United States and elsewhere. Under clause 3.2 of the AI terms, any MSA indemnity for infringement of intellectual property or other third party rights excludes generated AI output. Clause 3.1 assigns the customer the duty to identify and correct output errors before relying on a result. The March 2025 AI terms retain a link to their February predecessor, making the version boundary identifiable. This arrangement defines the treatment of AI output directly, but leaves customers without the MSA's infringement remedy for that output. Human review is the customer's contractual responsibility.
The consolidated agreement for the United States entity has a general cap of twelve months of fees paid or payable, a fivefold cap for specified confidentiality, legal compliance and indemnity claims, and exceptions including fraud and gross negligence. It also grants ordinary service intellectual property protection. However, the AI annex expressly excludes AI Services from that intellectual property indemnity and disclaims consequences of relying on generated output. Sidebar is excluded from the uptime undertaking and credits. The separate signup terms still state an INR 100 cap. The applicable order and agreement matter; none of these provisions establishes a warranty that an AI legal conclusion is correct.
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.
Contract creation and management connect to business systems and Microsoft Word. Word synchronizes edits, comments, redlines and versions with the contract workspace. Slack, email, Google Drive and Word are available across plans. Growth adds HubSpot, Claude, Copilot, DocuSign, Greenhouse, Zapier and Ashby; Enterprise includes Salesforce, REST API access and webhooks. Claude uses MCP to work with the repository, while the ChatGPT connection remains a beta. Pricing depends partly on integration complexity, and the pricing FAQ identifies Salesforce, HubSpot and Workday as integrations carrying additional cost. No iManage or NetDocuments connection is identified. The integration set primarily follows in house contracting and business system workflows.
More than 30 integrations are stated, each with its own page. Salesforce, HubSpot, Slack, Microsoft Word for desktop, Google Drive, DocuSign, Greenhouse, Google Forms, Jira and Zapier are all linked directly from the pricing page. VerifAI runs review inside Microsoft Word, and negotiation and redlining are described as working in Word, Slack or SpotDraft itself. One named customer credits the Word desktop editor with driving adoption. Single sign on covers Office 365, Google Workspace, Okta, Active Directory and custom SAML with zero touch provisioning. The connections center on corporate business systems; a native iManage or NetDocuments integration is not described.
Deployment Model and Data Residency
Where the software runs and where the data sits. Multi tenant cloud, single tenant, private deployment, on premises, and whether region of residence is a published option or an enterprise conversation.
AWS supplies part of Juro's infrastructure, and the responsible AI guidance places GPT processing on Microsoft Azure servers. Customers use the hosted contract workspace across more than 85 countries. An EU Data Act addendum addresses a separate contractual requirement around the service. That international reach does not itself establish where an individual customer's contract data resides. The deployment material underlying this assessment does not specify a choice of regions, a residency commitment, tenant architecture or a private installation. The distinction between platform infrastructure and the model processing route is identifiable, but the customer controlled boundary for hosting location remains unclear.
The DPA assigns primary personal data storage to the Netherlands for EU customers, the United States for customers there and India for Indian customers. It permits controlled support and implementation access from India for other customers. The public security description still offers an additional Middle East region and says regional personal data does not leave its location, creating a distinction that requires contractual clarification. VerifAI identifies Azure hosting alongside OpenAI and Anthropic models, while the subprocessor register adds other services. Their registered addresses are not processing regions. Consequently, whether every model call stays in the customer's primary storage country is left open.
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.
Privacy regulations, application security, network infrastructure, resilience, electronic signatures and responsible AI each get a section in Juro's trust center. Its security claims include SOC 2 Type 2, supported by a corresponding trust center topic and a certification badge. The platform also has a status page and a stated 99.9 percent uptime. Access to the explanatory articles does not require a form, email address or confidentiality agreement. The trust center index does not identify the auditor, attestation period, report date or tested scope. Nor does it give a route to the underlying report or details of a penetration test. The security claim is identifiable, while its independent verification details remain incomplete.
Four compliance marks, for ISO, GDPR, HIPAA and AICPA SOC 2, appear on SpotDraft's home, pricing and security pages. Its home page lists them as ISO 27001, SOC 2 Type II, GDPR and HIPAA, and the footer on every page reads "An ISO/IEC 27001:2022 Certified Company". SpotDraft's help center lists its SOC 2 Type 2 and ISO 27001 reports with download links, and a March 2023 blog post announced the SOC 2 Type 2 result. SpotDraft also links a separate trust center at trustcenter.spotdraft.com. Outside the reports themselves, no auditor, coverage period, report date or scope is published for the SOC 2. Independent penetration testers are said to be used, but none is named and no summary is published.
Model Supply Chain Disclosure
Which models sit underneath, whose they are, where they run, and whether the vendor commits to telling customers when that changes. A legal buyer inherits every dependency it cannot see.
Behind the AI tools sits OpenAI's GPT family, delivered through Microsoft on Azure servers. The responsible AI guidance therefore names a provider, a model family and a delivery route. The discussion also uses GPT 4 as an example of training scale, without confirming that version as the current production model. Juro gives no precise version and no commitment to notify customers of model changes. The trust center's infrastructure subprocessor list names Algolia, AWS and MongoDB but omits OpenAI and Microsoft. The named model route therefore does not resolve the complete chain of processors or how customers learn of a change.
VerifAI identifies OpenAI GPT 4 and Anthropic Claude and describes selecting models by task, with an Azure instance also mentioned. The broader subprocessor register names Azure, OpenAI and Anthropic for metadata extraction, contract questions and guideline review, and Pinecone for vector retrieval. It states that those AI providers receive no data when AI features are unused. The DPA adds thirty days of notice before a subprocessor addition or replacement, with a customer objection route. SpotDraft still gives no individual model versions, no full list of inference regions and no notice of changes within an existing provider. The single model nutrition label does not fully describe this wider supply chain.
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.
Monthly contract volume and integration complexity set the price, with unlimited users, workflows and templates across plans. Volume bands range from fewer than 20 contracts per month to more than 1,000. Slack, Google Drive and Companies House are included according to the pricing FAQ, while Salesforce, HubSpot and Workday incur additional cost. Quotes also reflect implementation complexity. Billing can use USD, GBP or EUR, and nonprofit discounts are available. No numeric subscription or implementation price is stated. The pricing page supplies the main drivers of a quote without a per seat rate or an independently calculable total. A previous demo month promotion is no longer part of the current pricing offer.
SpotDraft prices the CLM by users or contract volume and includes standard implementation, migration, integration setup and support in its published package. It does not publish a subscription rate or a separate allowance for each AI feature. The consolidated agreement permits additional charges for custom work outside the order's scope, requires advance payment and treats fees as a minimum commitment. Its AI modules can be purchased separately under the order. Indian stamp procurement has a distinct fee schedule and deposit under the India agreement, so stamping is not automatically part of an unlimited CLM price. VerifAI advertises a two week trial.
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.
Buyers span legal, HR, procurement, sales and finance teams. Legal establishes templates and workflow controls, allowing other departments to produce agreements within an approved process. Contract types include NDAs, employment agreements, master services agreements, statements of work, order forms, acquisition material and supplier agreements. The company describes customers in more than 85 countries and positions the platform for businesses completing roughly ten or more contracts each month. Its industry examples include software, food delivery and marketplaces. The central setting is an in house contract operation. A dedicated law firm, government or court workflow is not identified, and the product does not function as a primary legal research service.
The core audience is corporate legal departments working with sales, finance, HR and procurement colleagues. Industry examples span software, HR technology, education technology, health technology and financial technology. The consolidated agreement limits use to the customer's own business and prohibits operating the service for third party benefit, which is relevant to a firm considering client delivery. VerifAI supports contract review through configurable playbooks, while Sidebar adds broader legal and operational assistance. Sidebar's prohibition on sensitive personal and financial information limits the material that can be submitted. No complete list of supported legal jurisdictions or excluded contract issues accompanies the product offering.
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?
Customer data may not train a model for anyone else under Juro's AI terms. A model used only to provide services to that same customer is permitted, and Juro must erase it within 60 days after the agreement ends. The responsible AI guidance also rejects training foundational models with customer data. The contract's limited customer specific exception is therefore distinct from shared model training, and the deletion deadline applies to that derived model.
The consolidated AI annex expressly prohibits using customer data as a training dataset for machine learning, reinforcement learning or model training. It permits usage analytics and storage of generated information for debugging a reported issue, but those permissions do not override the training exclusion. The DPA separately bars training on customer personal data. These protections apply through the relevant signed agreement and order, including the consolidated Sidebar terms; the product nutrition label alone is not the basis for the contractual commitment.
Prompt and Output Retention
How long does the product keep what a lawyer typed, and can that be set to zero?
A customer specific model must be erased within 60 days after termination under Juro's AI terms. That deadline governs the model rather than the contracts, prompts or generated output used in the service. Customers can exercise portability and erasure rights under the privacy framework. Juro gives no content retention window and no customer controlled retention setting, so the lifetime of operational AI content sits apart from the model deletion commitment.
VerifAI offers deletion on demand, and the DPA provides customer directed processing duration and deletion. On termination or request, personal data is returned within thirty days, with copies deleted within ninety days of return or as directed, subject to stated exceptions. The consolidated agreement separately permits retrieval of contracts and Sidebar chats, knowledge files and tables after termination. Generated information may also be stored to debug reported issues. No universal zero retention setting or provider log schedule is established.
Ethical Walls and Matter Segregation
Does retrieval respect the firm’s ethical walls, or can the model read across them?
Legal, sales, HR, procurement and finance can work in Juro's contract workspace. The AI terms prevent a customer specific model from serving another customer and distinguish one user's output from similar output produced for others. Those contractual boundaries do not explain permission enforcement within the repository or retrieval across teams. Juro describes no technical separation model and no controls for ethical walls between matters.
The platform separates customer data within shared infrastructure, uses a distinct encryption key per contract and scopes permissions by department, entity and contract type. The DPA additionally promises isolation between Sidebar customer environments and limits support access through roles, least privilege and logging. The result is a tenant and organizational separation model for an in house buyer. SpotDraft does not say that every model retrieval rechecks the same permissions, and the design offers no law firm ethical wall spanning multiple client matters.
Third Party Request and Subpoena Notice
If someone subpoenas the vendor for a firm’s data, does the firm hear about it first?
Juro's data processing agreement requires notice to the customer of government access requests under Appendix 3, paragraph 4.4(k). The commitment operates through the Standard Contractual Clauses where the corresponding transfer law applies. Notification to individual data subjects remains the customer's responsibility. An equivalent commitment is not identified in the general terms or CCPA appendix. Requests from a third party claiming to control customer personal data are directed to the customer. No transparency report or request volume accompanies these provisions.
The consolidated agreement promises advance notice of compelled confidential information disclosure when reasonably practicable and legally allowed. The DPA further requires notice within 72 hours of awareness of government access demands or relevant regulatory proceedings, with consultation and efforts to allow a protective response before disclosure. Legal prohibitions and compulsion remain exceptions. These are customer notification commitments rather than a published transparency report, and they do not guarantee that the customer can prevent a lawful disclosure.
Primary Law Corpus Provenance
Where does the law in this product come from, and does the vendor have the right to use it?
Operator answers questions from the customer's contract repository and connects responses to its sources. AI Extract captures data from signed agreements in that same customer supplied collection. The underlying model route is separately identified as OpenAI's GPT family through Microsoft Azure. Juro is not a primary law research corpus. Rights to the foundation model's training material and an update schedule for that material are not established by the repository's customer supplied provenance.
VerifAI uses the customer's contracts and playbooks, supplemented by guides developed with legal professionals. Sidebar also describes monitoring official regulatory sources, and the AI terms acknowledge the use of information from the internet. These statements do not enumerate a licensed legal authority collection, identify the authors of every guide or explain rights for all external material used in responses. A reference back to a supplied agreement establishes its local source without establishing the provenance of the model's wider legal knowledge.
Good Law Verification
Does the product tell you when the authority it just cited has been overruled?
Contracts belonging to the customer, rather than a primary law collection, are what Juro works with. It has no identified citator, treatment signal or process for checking the subsequent history of a legal authority. Operator's citations connect repository answers to contract sources, which serves a different purpose from determining whether a case remains good law. The AI terms acknowledge that generated output can misstate laws and legal authorities and require human correction before reliance.
SpotDraft's home, pricing and security pages describe no citator, treatment signal or currency check, and do not say whether legal authority is reviewed for later history. The platform manages a customer's own contracts rather than retrieving case law or legislation, so a citator is not part of what it sells. Sidebar is described as helping users stay ahead of regulatory change with AI agents. That concerns the currency of regulation rather than the standing of cited authority, and no source or verification method is published for it.
Refusal and Uncertainty Behavior
What does the product do when the answer is not in the corpus?
Juro warns that generative AI can produce errors and requires a person to review results before reliance. Playbooks provide context and constraints for contract tasks, but no defined abstention path or confidence indicator explains how the system responds when an answer lacks support. The responsible AI guidance treats accuracy as an ongoing professional review responsibility. Juro thus pairs a warning with a human control, but does not say when the AI itself declines a request.
VerifAI explains whether contract provisions meet a selected guideline and presents proposed edits for review. Its terms warn that generated content can be unreliable and require independent verification. Neither statement describes a model refusing a question when authority is missing, flagging a specific uncertainty level or stopping an agent because the playbook does not cover a request. Contractual restrictions on high risk use are customer obligations, not demonstrated refusal behavior implemented by the system.
Fabricated Citation Record
Does a public court record exist addressing fabricated or hallucinated legal citations in output from this product?
No identified court decision or disciplinary record names Juro or Juro Online Limited in connection with fabricated citations. Damien Charlotin's AI Hallucination Cases database and the sanctions material underlying this assessment contain no attributable incident. The absence of an incident gives no error rate and says nothing about whether generated authorities are reliable. Juro primarily supports commercial contracting inside companies, a setting in which output is less likely to become part of a court filing.
The Charlotin litigation database contained no entry naming SpotDraft, VerifAI or Draftspotting in the AI tool field or other case information when reviewed on October 16, 2026. Contract workflow software is less likely to be identified in a filed legal brief than a research assistant, although Sidebar now extends the product's legal assistance. This negative result does not measure hallucination frequency or establish that every generated reference is accurate; it records the absence of a named case in that database.
Bar Guidance Alignment
Has the vendor engaged in public with the ethics opinions its buyers are bound by?
Lawyers must protect professional confidentiality when using AI, Juro recognizes. Its responsible AI guidance connects that duty to restrictions on training with customer data, while making a similar confidentiality point for HR teams. It does not connect these controls to a named ethics opinion or regulator's requirements. ABA Formal Opinion 512, state bar guidance, and Solicitors Regulation Authority or Law Society materials are not identified as the governing framework.
SpotDraft's home page names two ethics guidance documents from two jurisdictions. One is the California State Bar's Practical Guidance for the Use of Generative Artificial Intelligence in the Practice of Law, dated November 2023. The other is the American Bar Association's Formal Opinion 512 on generative AI, dated July 2024. Both are given with issuer and date. SpotDraft says its AI is designed around both, for responsible and secure contracting. The statement does not map which duty each product control meets, and no other jurisdiction's guidance is addressed.
Billing and Fee Posture
Does the vendor address what happens to the bill when the work takes an hour instead of six?
Reduced contract work, including faster signing, fewer manual reviews and time saved per agreement, runs through the customer examples. The commercial focus is an in house operation where legal and business teams share the platform. Those efficiency claims say nothing about how a law firm should bill a client for AI assisted work. No matter based fee record, disclosure procedure or guidance on passing time savings through to clients accompanies the contracting workflow.
SpotDraft sells contract operations to corporate departments, and the consolidated service terms restrict use to the customer's own business rather than delivery for third party benefit. Its cost and time savings claims concern that internal workflow. User or contract volume charges are software procurement terms, not a system for recording a lawyer's billable contribution to a client matter. No legal client expense allocation, fee disclosure template or professional billing guidance accompanies the AI features.
Outside Counsel Guideline Readiness
Can a firm get this vendor through a client’s AI clause without a bespoke negotiation?
Juro identifies infrastructure, analytics, communications and payment subprocessors, including AWS, MongoDB, Algolia, Stripe and HubSpot. Customer initiated integrations include Salesforce, Slack and Google. The contractual package includes a data processing agreement, an EU Data Act addendum and separate AI terms, with an impact assessment template available for AI Assistant. The trust center list does not include the OpenAI and Microsoft route described in the responsible AI guidance. No client consent or notification pack accompanies this material.
The published register identifies hosting, execution, support and AI suppliers, including Google Cloud, AWS, Azure, OpenAI, Anthropic and Pinecone. Sidebar has additional named providers. The DPA requires thirty days of advance notice for changes and allows objections on data protection grounds. Together with contractual training restrictions, these materials can inform an organization's AI procurement approval. They do not supply a ready made outside counsel consent package, and the register's company addresses should not be mistaken for where each supplier processes contract data.
Court Disclosure Support
If a judge’s standing order requires an AI disclosure, can the product produce one?
Commercial contracts, not court filings, are what Juro manages. Operator provides source citations for repository answers, and the AI terms prohibit representing generated output as human authored. These features support a limited record of sources and a contractual disclosure boundary. They do not create a court certification, standing order workflow or export that records the model's role and a lawyer's verification. A dedicated litigation disclosure process is not identified.
Contract audit logs retain creation, signing and version changes. The consolidated agreement also makes Sidebar chats, uploaded knowledge files and generated tables retrievable for thirty days after termination, subject to the applicable order. Its AI terms prohibit presenting generated material as human authored. Those provisions can preserve part of a usage history, but no court specific export joins model identity, prompts, source verification and the lawyer's approval into a complete filing declaration. The available record therefore remains partial.
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 Juro if
- You want the agreement in front of you before the demo. Juro publishes master services agreements for United States and non United States customers and standalone AI terms carrying a last updated date of 11 March 2025, with the superseded version still linked. A data processing agreement, an EU Data Act addendum, API terms and evaluation terms are all dated downloadable documents too. Clause 3.2 of those AI terms states where the indemnity stops, which is at AI output.
- You need the training prohibition in a term rather than on a page. Clause 1.2 of Juro's AI terms states that Juro may not use customer data as training data for any AI model, with a single carve out for a customer specific model used only to serve that same customer, and requires any such model to be erased within 60 days of termination for any reason.
- A client asks whose model reads its contracts. Juro states openly that its AI runs on OpenAI's GPT model provided by Microsoft on Azure servers. That one sentence names the model family, the provider, the delivery route and where inference happens. It also publishes a subprocessor list and ten trust center articles readable with no form, email or NDA.
Choose SpotDraft if
- Your data has to stay in a named region. SpotDraft states that customer data is logically separated within a secure multi tenant infrastructure. Personal data is stored within a selected region across the United States, the European Union, India and the Middle East, and is not transmitted outside those locations. Primary and backup servers sit on Google Cloud Platform in the Netherlands, with per contract encryption keys in HashiCorp Vault.
- Your risk committee wants the ethics guidance named. SpotDraft states that its AI features are designed with attention to the California State Bar's Practical Guidance for the Use of Generative Artificial Intelligence in the Practice of Law of November 2023 and to the American Bar Association's Formal Opinion 512 of July 2024, both given with issuing body and date.
- You do not want implementation quoted separately after signature. SpotDraft states that in house implementation is always included, covering workflow and integration setup and migration of legacy contracts, with no extra fees and no outsourcing. It publishes a week one to week six rollout sequence, and includes a dedicated customer success manager and support at no additional cost.
In summary
Juro
Juro is an end to end contract lifecycle platform for in house teams, built so that colleagues outside legal can create and complete contracts inside templates and workflows legal controls. Its AI covers AI Review for triage and redlining, AI Extract for pulling data from executed contracts, and an assistant. Operator is a conversational agent over the repository that cites its sources. According to the AI Legal Index, Juro's most specific published material is contractual: clause 1.2 of its AI terms bars using customer data to train any model except a customer specific one serving only that customer, erased within 60 days of termination. It publishes no accuracy measurement, no retention period and no statement of tenancy or region.
SpotDraft
SpotDraft is a contract lifecycle platform for in house legal teams. It covers template driven creation, conditional approval workflows, negotiation and redlining in Word, and native eSignature. Its repository extracts more than a thousand metadata types, and its AI runs through VerifAI review, Intake and the Sidebar agents. According to the AI Legal Index, SpotDraft's case rests on deployment and residency. Customer data is logically separated within a multi tenant infrastructure, and personal data is held in a customer selected region across the United States, the European Union, India and the Middle East and stated not to travel outside it. Servers run on Google Cloud in the Netherlands, with per contract keys in HashiCorp Vault. It publishes no accuracy figure, no model provider and no training position.
Questions buyers ask
Juro vs SpotDraft: which is better for an in house legal team?
They differ by disclosure rather than function. Juro publishes its agreements, its AI terms and the model layer behind the product. SpotDraft publishes where data is stored and processed, names two ethics guidance documents, and includes implementation at no extra cost. Neither publishes an accuracy measurement or a rate. From the AI Legal Index, based on each vendor's own published materials as of September 3, 2026. No vendor pays for placement.
Does Juro train its AI on customer contracts?
No, and the prohibition sits in the agreement rather than on a policy page. Clause 1.2 of Juro's published AI terms states that Juro may not use customer data as training data for any AI model, carving out only a customer specific model used solely to serve that same customer, which must be erased within 60 days of termination for any reason. Its responsible AI page states the same commitment in summary form, that customer data is not used to train foundational models. From the AI Legal Index, based on each vendor's own published materials as of September 3, 2026. No vendor pays for placement.
Where does SpotDraft store contract data?
SpotDraft states that personal data is stored within a region the customer selects across the United States, the European Union, India and the Middle East, and is not transmitted outside those locations. Primary and backup servers run on Google Cloud Platform in the Netherlands, and encryption is AES-256 at rest with a unique key per contract held in HashiCorp Vault backed by Google Cloud KMS. The regional commitment is written in terms of personal data, and nothing published confirms that contract content follows the same rule. From the AI Legal Index, based on each vendor's own published materials as of September 3, 2026. No vendor pays for placement.
Do either name the AI models they use?
Juro does. It names OpenAI's GPT model provided by Microsoft on Azure servers, though it gives no version and publishes no commitment to notify customers of a change. Neither OpenAI nor Microsoft appears on Juro's published subprocessor list, which names Algolia, AWS and MongoDB as infrastructure. SpotDraft names no model, provider or architecture at all. From the AI Legal Index, based on each vendor's own published materials as of September 3, 2026. No vendor pays for placement.
What do Juro and SpotDraft both leave unpublished?
Neither publishes an accuracy measurement for contract review, whether an error rate, a test set, a benchmark or an evaluation. Neither states a retention period for contracts, prompts or generated output. Neither publishes a rate, although both publish the unit of charge and the structure around it. And neither addresses legal professional privilege or work product, which matters because both platforms hold the whole contract estate for a legal department. From the AI Legal Index, based on each vendor's own published materials as of September 3, 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.
Juro's AI terms say plainly where responsibility for AI output sits. Clause 3.2 states that the indemnity in the master services agreement does not apply to AI output, and clause 3.1 puts the duty to identify and correct errors on the customer. So the allocation of loss for wrong output is published, dated and versioned before any sales conversation. On SpotDraft, the published Terms of Use cap total liability at one hundred Indian rupees. The same document contemplates further terms where other services are purchased, so an enterprise buyer may sign a master agreement that is not published. Its site footer states ISO/IEC 27001:2022. Neither vendor reviewed this page.