ContractKen vs ThoughtRiver: how they compare in 2026
ThoughtRiver is better aligned with a legal department coordinating review volume, portfolio analysis and shared access across many users. Its annual packages distinguish workloads below and above a monthly contract threshold, and its evaluation material describes classification tests with separate precision and recall measures. ContractKen is a more direct fit for the individual lawyer or firm team drafting and negotiating in Word, particularly where detailed fallback positions and local masking are useful. The numerical performance claims should not be treated as interchangeable guarantees: ThoughtRiver's classification measures do not establish every generated answer, and ContractKen's architecture description is not a comparable universal accuracy result. Both allow specified improvement or training uses in their agreements, while their deployment and commercial units differ substantially.
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.
Models classify clauses, compare negotiation positions, explain deviations and generate contract drafts or proposed edits. The current pipeline combines language inference, entity recognition, retrieval from approved material and generative models selected for the task. These functions are central to the review and drafting work. The Word application also retains useful nonmodel components: a shared clause library, playbook administration, formatting checks and reference validation. Users can organize approved language and maintain negotiation standards alongside the AI analysis. ContractKen thus joins model driven legal work to an established editing and knowledge workflow, rather than treating every formatting or storage operation as artificial intelligence.
Machine learning is the mechanism the customer pays for. A model reads the contract, answers the Lexible question set, scores the answers against a playbook and produces the risk rating, issues list and suggested redlines. Remove the models and there is no review, no issues list and no redlines, only a document store. The vendor has sold this as an AI product since 2016 and describes the current engine as a proprietary generative model on several large language models.
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.
Review findings connect to the relevant contract passage, and generated explanations draw on the customer's playbooks, clause library and prior agreements. Retrieval preserves section structure, definitions and cross references instead of splitting text without that context. Drafted clauses identify whether they came from a precedent, approved library language or a playbook requirement. Low confidence suggestions are flagged for review. The models undergo internal classification tests before deployment, but current numerical results, a reproducible test set and error rates for generated legal conclusions are not specified. A research result about similar legal models is separate from ContractKen's production accuracy.
Lexible uses lawyer labeled concepts to identify contract provisions and assess them against a playbook. Its accuracy explanation reports 97% F1, 96% precision and 97% recall, with testing three times weekly against 750,000 verified data points. Legal specialists verify training examples, and sampling covers different contract types and domains. These figures concern classification and do not measure every generated redline or answer. The underlying dataset, current model version and detailed failure results are not available for independent reproduction. The terms acknowledge that training can reduce accuracy, especially with incorrect labels, and that mistakes remain possible.
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.
Playbook Driven Review checks preferred, fallback and walkaway positions; Comprehensive Review provides an alternative when no organizational playbook exists. Risk sensitivity can be adjusted, and the playbook framework distinguishes deviations that require escalation to a partner. Reviewers navigate to the exact clause, inspect an explanation and approve, modify or reject a proposed change. No edit enters the agreement without explicit human approval. Drafts receive a playbook compliance report before use, and administrators control who can change the playbook itself. These controls separate automatic analysis from the lawyer's decision about the final language and the route for an unacceptable negotiating position.
The platform reads an agreement, identifies issues against a configured playbook and presents suggested redlines in Word. Users can share issues and changes with colleagues, while Lexible Assistant supports questions and remediation. The customer sets its risk policies and remains responsible for decisions based on the resulting analysis, including automated decisions. Terms expressly warn that output is not legal advice and may contain mistakes. These features provide a review surface and organizational standards. The system does not specify when it can approve an agreement without review or how it must escalate and correct an incorrect conclusion.
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.
Named practitioners describe using ContractKen in contract work. Aseet Patel, a lawyer and legal technology founder, uses it for agreements and term sheets before sending work to outside counsel for final review. Sharon Zachariah, an innovation lawyer and former partner, describes using it after drafting or reviewing client contracts. These accounts identify people and actual tasks, but give no dated baseline or measurement method. Separate drafting illustrations promise substantial reductions in time to a first draft, and the review calculator estimates savings from assumed workloads. Those illustrations do not establish measured outcomes for either named practitioner or the current AI pipeline.
A named customer gives figures, without a date or method. Joe Stephenson, partner and head of technology at Shoosmiths, is quoted on the demo page saying the platform reviewed complex supply agreements in under three minutes at above ninety percent accuracy, where lawyers took four hours at 86 percent. A second, unnamed customer calls ThoughtRiver a clear winner for reviewing its appointments. No date, sample size or method accompanies the Shoosmiths figures.
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 security commitments treat content as confidential work product, restrict staff access to necessary support under confidentiality obligations and separate customer tenants with individual encryption keys. The Moderation Layer masks sensitive text locally, offers a preview and validates the sanitized version before external model processing. Public or shared model training is excluded in the security FAQ, while the architecture also describes controlled feedback learning and private customization. The general terms reserve content review for service improvement. Retention and purge controls exist, but internal matter walls and a consistent contractual training restriction are not established across these different surfaces.
The agreement treats customer contract data as confidential and restricts disclosure by staff and subcontractors, with confidentiality continuing after termination. Customer accounts have separate encryption keys and segregated contract data. Anonymization removes sensitive information before model processing. The same agreement grants continuing use of training data and anonymized contract text, and the DPA includes machine learning training among processing purposes. This is permission to reuse specified material, rather than a promise never to train. Internal matter walls, direct privilege handling and a complete account of every underlying language model provider are not specified.
UPL and Professional Responsibility Posture
Whether the vendor is clear that it supplies a tool rather than legal advice, who its audience is, and how it addresses unauthorized practice of law, competence and supervision duties, and jurisdiction limits. ABA Formal Opinion 512 is the reference point. Where the advice line is not the duty a product raises, the axis is read through the nearest professional duty it does raise: judicial conduct rules and the reviewing duty for products sold only to courts, and the duty to bill for time actually spent for products that draft time entries.
The terms distinguish ContractKen from a law firm or attorney and state that its information is not legal advice. They direct people needing advice to a licensed lawyer and do not create a professional relationship with the vendor. The product reinforces supervision through source inspection, playbook controls and required approval of generated changes. Its confidentiality explanation also engages with lawyers' duties under ABA Formal Opinion 512. Standard clauses include common law jurisdiction notes, but that coverage does not define a complete set of professional practice limits. The trial audience includes people who are not lawyers as well as legal teams.
The customer terms expressly say that output is not legal advice and must not be relied on as advice. They also warn that predictive technology makes mistakes and that incorrect training can worsen accuracy. Customers remain responsible for results, conclusions and automated or manual decisions. The license allows both internal business work and integrated managed services for legal clients, so law firms and other business users share that responsibility. Product review surfaces support human checking, but jurisdiction specific practice limits and an explicit treatment of lawyers' competence and supervision duties are not set out.
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.
ContractKen describes a controlled model update process using feedback from accepted, modified and rejected suggestions. Those signals are aggregated and reviewed before supervised retraining, rather than changing the model instantly after each user action. Updated models face held out tests using precision, recall and F1 measures; performance must remain stable or improve before release. Administrators can also restrict sensitive information through redaction rules. This is a concrete development and release framework. A named person accountable for AI governance, current test results and findings about uneven performance across jurisdictions, contract types or groups are not specified.
ThoughtRiver describes lawyer verification of training examples, annotation that accounts for contract context and sampling across contract types and legal domains. Its evaluation program runs three times each week against 750,000 verified data points and reports separate precision, recall and F1 measures. Customers can set playbooks for their own risks and review proposed changes before further use. These are concrete training and evaluation practices. The description does not identify an accountable AI governance owner, a mandatory release approval threshold or findings about uneven output across jurisdictions or populations. Its regulatory position is separate from those operating tests.
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.
Administrators can request an immediate purge of contracts, prompts, logs and backups. Operational logs normally remain for 30 days, with a contractual option to set that period to zero. Enterprise controls include single sign on, administrator multifactor authentication and permissions for playbook uploads, draft review and data export. Staff access is restricted and logged. The security FAQ states continuous monitoring and material incident notification within 72 hours or a faster legal deadline. A complete current subprocessor register and the DPA require a request, and the default retention period for application prompts and generated output is not separately stated.
The DPA names infrastructure, authentication, document conversion, email and support providers, and requires equivalent safeguards for subprocessors. Personal data breaches must be reported without undue delay, with cooperation in mitigation. Customer data is encrypted in transit and at rest, separated by account and protected by a dedicated encryption key. Authentication supports single sign on and multifactor controls. At termination, personal data must be deleted or returned, with backup and legal exceptions. A defined in use retention period for prompts, a backup deletion deadline and an explicit list of the language models processing contract text are not specified.
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 public terms disclaim accuracy, completeness and noninfringement warranties and cap aggregate liability at the greater of $100 or fees paid in the preceding 12 months. Exceptions concern death, property damage or injury from gross negligence, and fraud or fraudulent misrepresentation. Customers indemnify ContractKen for specified claims. The security FAQ offers a stronger position, including an IP indemnity, performance warranty, cyber and professional insurance and different liability exceptions. Paid customers may instead have an Enterprise Services Agreement, whose text is not available here. Those stronger commitments therefore do not clearly replace the default terms for every subscriber.
ThoughtRiver indemnifies customers against claims that permitted use of the platform infringes third party copyright, subject to control of the defense and exclusions for unapproved combinations. It can replace or modify the platform to address infringement. General liability is limited to the order term's fees, or £1,000 during a trial, with the copyright indemnity excluded from that cap. Death or injury caused by negligence, fraud and other legally unavoidable liabilities also remain outside the limitation. Output errors carry no warranty or accepted liability, so the software indemnity does not amount to a guarantee of correct legal analysis.
Practice Systems Integration Depth
How deeply the product reaches into the systems legal work already lives in: document management such as iManage and NetDocuments, Word and Outlook, contract lifecycle management, matter management, e-billing, and court filing systems.
The Word add in reads the active agreement chosen by the user, rather than scanning other files or email. Review findings navigate to the underlying clause. Approved suggestions return through Office.js as ordinary tracked changes with explanatory comments, and the privacy mapping restores original names and values. Users supply their playbooks, precedents and clause language, then review output within the same document. Standard setup requires Microsoft Word and includes onboarding support. Integrations with systems such as iManage belong to the separately quoted custom workflow offer; they are distinct from the detailed, standard Word editing connection.
Agreements enter through email, file upload or a Microsoft Word add in, with issues and redlines available for team collaboration. Microsoft 365 supports work in Word, Outlook and Teams, while Power BI receives contract data for analysis. HighQ and iManage are named connections for legal projects and document workflows. The platform also exposes the same public APIs used by its web application and follows the OpenAPI specification. These connections have identifiable purposes, but per connector field mappings, authentication setup, permission inheritance and error recovery steps are not described in the integration overview.
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.
The standard service combines a Word add in with a hosted analysis pipeline. Sensitive entities can be replaced locally before sanitized text reaches ContractKen's servers and external AI providers. Primary hosting is on AWS, with European Union or United States residency available on request. Regulated teams can obtain private inference in their own Azure tenant or virtual private cloud, and enterprise customers can supply an endpoint. These options distinguish local preparation from remote analysis. Exact cloud regions, the location of every provider's processing and the configuration or price differences between residency choices are not specified.
Region and processing location are stated, and tenancy is described as segregation. The security page says data resides in Azure data centers managed and secured by ThoughtRiver in a single specified region, and the subprocessor table places Azure hosting in the United Kingdom. Each customer's data is held separately under its own key. Whether a customer may choose a region other than the UK, and whether infrastructure is shared, are not stated, and where the language models run inference is not addressed.
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.
ContractKen describes SOC 2 Type II certification as in progress and its controls as aligned with ISO 27001 and ISO 27701. The security FAQ also mentions a report under a nondisclosure agreement, without identifying a completed audit period or service scope. Encryption, tenant isolation, administrator authentication, annual independent penetration testing and customer audit access are separate operating controls. Penetration test summaries and a security paper are available on request. Its own certification remains unfinished, without a dated independent audit, a named auditor or a defined assurance scope covering the ContractKen service.
ISO 27001 certification is stated and identified by number, without a route to the report. The security page gives certificate number 21188-ISMS-001 alongside encryption at rest and in transit. No certifying body, coverage period or route to the certificate or any SOC report is published, and no trust center 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.
The architecture identifies DeBERTa for clause classification and natural language inference, plus spaCy and custom entity types for recognition. Generative models handle explanations and edits after retrieval from approved material. The security FAQ names OpenAI and Google as example API providers, while enterprise customers can bring their own endpoint. Routing selects models by task and can change as performance improves. Exact generative model families, versions and processing regions are not specified. The 30 day notice commitment concerns material changes to data handling or service levels, rather than every substitution of a model in the pipeline.
Lexible combines a proprietary legal model with general purpose language models for contract interpretation and generated responses. ThoughtRiver describes supervised learning from expert labeled concepts, legal validation of examples and repeated evaluation. An anonymization layer protects sensitive text before analysis, and a private model option is offered. The hosting environment is Microsoft Azure. External foundation model providers, versions, routing decisions and processing locations for each model are not named. The subprocessor register covers infrastructure and support functions but does not resolve the generative model chain or commit to notice of each model substitution.
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.
The standard Word subscription costs $149 per user per month on monthly billing or $119 with annual billing. It has no minimum seat requirement and includes unlimited review and redlining, a clause library, playbooks, formatting, onboarding and training. A guided trial lasts 14 days, and cancellations or seat reductions can receive a prorated refund. Custom AI and workflow projects use a quoted proof of concept lasting eight to 12 weeks, including usage and customization. Standard installation is included; custom implementation support is also described as included, but the project fee and enterprise deployment price are not stated.
Professional starts at £15,000 a year in the United Kingdom for teams reviewing fewer than 50 contracts a month; Enterprise starts at £30,000 for higher volume. Outside the United Kingdom, the page displays starting rates of $20,000 and $35,000 respectively. Both list unlimited users, a private database instance, Word access, Lexible Assistant, single sign on and dedicated success and professional services. A trial lasts 28 days. The agreement bills order fees in advance and permits separate professional services charges. The published starting amounts therefore do not settle every implementation charge, volume adjustment or negotiated order term.
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.
Law firms and internal legal teams can review commercial agreements, draft from prior transactions and maintain shared negotiation language. Examples include nondisclosure, service, software, employment, supply, lease and acquisition agreements. Prebuilt clauses cover common law jurisdictions including the United States, United Kingdom and Australia, with jurisdiction notes and tags for custom language. Separate playbooks can serve different teams, contract types or jurisdictions. Multilingual review is described broadly, but a tested language matrix and unsupported practice boundaries are not specified. The product centers on transactional drafting and negotiation; its clause guidance does not establish a general legal research or litigation service.
Corporate legal, procurement and business teams use ThoughtRiver to review incoming agreements, and the license also allows managed services delivered to legal clients. Configurable playbooks apply the customer's negotiation standards to recurring contracts. Professional and Enterprise packages distinguish teams below and above 50 reviews per month, with unlimited users in both. Named examples include complex supply agreements and professional appointments. The current site gives different totals for its trained concept library on different pages, so that total does not define a reliable coverage boundary. A complete jurisdiction matrix and unsupported contract types are not stated.
The 12 legal signals, side by side
Recorded rather than graded. These are the questions a practitioner has to answer before a tool touches a client matter, and the answers are taken from public material only.
Client Data in Training
Can material a lawyer puts into this product be used to train a model?
Section 8 permits review or editing of content for "operating and improving the Service" without a model training exclusion. The security FAQ excludes training public or shared models, while the architecture describes aggregated lawyer feedback used in controlled retraining and private customization. Another statement says no customer data trains models. The Enterprise Services Agreement may supersede the general terms, but its training clause is not available in the public agreement.
The agreement licenses training data and anonymized contract text for use in the platform by ThoughtRiver and other customers, while excluding customer risk policies from that licensed material. Its DPA expressly lists machine learning training and feature enhancement among processing purposes. These permissions continue beyond the subscription and are separate from customer ownership of uploaded contracts. The public agreement does not provide a general option to exclude those uses.
Prompt and Output Retention
How long does the product keep what a lawyer typed, and can that be set to zero?
Administrators may request immediate deletion of contracts, prompts, logs and backups. The security FAQ gives operational logs a 30 day default and allows that period to be set contractually to zero. It does not separately define a default window for application prompts or generated answers. The privacy policy's period after account termination concerns personal information and does not establish how long AI interactions remain during active use.
The DPA requires deletion or return of personal data at termination, except for backup copies and legally required retention. Trial data is deleted when a customer properly ends the trial. The general agreement separately preserves rights over licensed training data and anonymized contract text. None of these provisions gives prompts or generated answers a defined retention period during an active subscription, and the backup exception has no stated deletion deadline.
Ethical Walls and Matter Segregation
Does retrieval respect the firm’s ethical walls, or can the model read across them?
Customer tenants are logically separated and use individual encryption keys. Enterprise roles control who uploads playbooks, reviews drafts or exports information, and shared precedents are available to authorized team members. These controls separate organizations and administrative activities. A permission model that divides opposing matters within one firm, or carries a connected document system's ethical walls into retrieval at query time, is not specified in the standard Word workflow.
Contract text and derived data are separated from other customer accounts and encrypted using a key specific to each account. Organization access can use single sign on with multifactor authentication. These are organization level protections. Although iManage and HighQ are named integration options, a mechanism that inherits their per user permissions during retrieval or separates opposing matters within one customer environment is not specified.
Third Party Request and Subpoena Notice
If someone subpoenas the vendor for a firm’s data, does the firm hear about it first?
The terms allow ContractKen to preserve or share information when reasonably necessary for legal process, law enforcement and protection of rights or safety. The privacy policy also permits disclosures in response to legal requirements. Neither sets out advance customer notice for a subpoena, a process for challenging a request or a transparency report. The separate security incident notice commitment concerns breaches and does not supply a legal demand notification policy.
When law requires ThoughtRiver to process personal data contrary to customer instructions, the DPA requires notice to the customer unless law prohibits it. That protection covers compelled processing of personal information in uploaded material. The agreement also permits legally required disclosure of confidential information, without a broader notice procedure for nonpersonal material. A public report of government demands and the response to them is not specified.
Primary Law Corpus Provenance
Where does the law in this product come from, and does the vendor have the right to use it?
Review and drafting use the customer's playbooks, clause library and prior agreements. ContractKen also supplies a library of commercial clauses and describes bundled language as authored internally, licensed or drawn from public domain material. Individual external sources and their rights are not listed, nor is an update schedule for a primary law corpus. Jurisdiction notes and refreshed sample language support drafting without establishing a comprehensive collection of legal authority.
The engine applies a lawyer developed concept library to customer agreements and risk policies. ThoughtRiver describes millions of labeled concepts and legal verification of training examples, while the agreement defines training data and anonymized contract text available for broader platform use. These identify the main categories behind contract interpretation. The underlying external legal collections, their licenses and the refresh schedule for any legal authority are not specified.
Good Law Verification
Does the product tell you when the authority it just cited has been overruled?
ContractKen can flag a precedent that no longer matches the team's current playbook and replace language from the approved clause library. It also traces generated clauses to their inputs. These checks concern contract standards and provenance, rather than the subsequent treatment of a judicial decision. A commercial citator connection, a method for checking whether authority has been overruled or an inline legal status warning is not specified.
ThoughtRiver's review concentrates on contract provisions, risk policies, proposed edits and obligations in customer agreements. A reviewer can work through issues in that agreement and use the resulting information for negotiation or reporting. The stated workflow does not include a licensed citator or a method for checking the subsequent treatment of a court decision. Classification accuracy and contract risk ratings do not supply that authority status check.
Refusal and Uncertainty Behavior
What does the product do when the answer is not in the corpus?
The interface flags low confidence suggestions and gives reviewers supporting sources, explanations and control over proposed edits. Playbook thresholds identify unacceptable negotiating positions and can send those issues to a partner. That escalation is about contract risk rather than unsupported answers. An explicit refusal response when retrieval cannot support a generated conclusion, or a published demonstration of the system declining such a question, is not described.
Lexible flags contract risks, produces proposed changes and lets a reviewer inspect issues against a playbook. Its precision, recall and F1 figures describe aggregate classification performance rather than a confidence indicator attached to an individual answer. The terms warn that predictive output can be wrong. A stated no answer response, an abstention threshold or a route triggered specifically by insufficient support for a conclusion is not described.
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 has no entry naming ContractKen as the tool behind fabricated legal citations as of October 15, 2026. Its stated work centers on contract drafting, clause review and proposed Word edits, with the lawyer approving changes. This court record concerns one specific kind of incident; it does not establish an error rate for those tasks or validate the accuracy of the current models.
The AI Hallucination Cases database has no tool entry naming ThoughtRiver or Lexible as of October 15, 2026. The product's described work involves contract provisions, risk review, redlining and questions about agreements rather than preparing a litigation brief. That workflow context is separate from the court record: the absence of a named entry does not measure classification errors or establish that a generated answer is correct.
Bar Guidance Alignment
Has the vendor engaged in public with the ethics opinions its buyers are bound by?
The Moderation Layer explanation addresses ABA Formal Opinion 512 and connects its confidentiality duties to local masking, user previews and logs of external AI requests. It discusses informed consent and also references several state bars, the SRA and the CCBE. It does not map the product to named opinions in multiple jurisdictions. These statements explain ContractKen's approach; they do not guarantee privilege or establish that every deployment satisfies a lawyer's duties.
The contract terms distinguish software output from legal advice and leave responsibility for decisions with the customer. They allow managed services for legal clients, and product controls let reviewers apply organizational playbooks and inspect proposed changes. These are general responsibilities and workflow controls. A named ethics opinion or jurisdiction specific bar rule governing a lawyer's use of Lexible is not identified in that position.
Billing and Fee Posture
Does the vendor address what happens to the bill when the work takes an hour instead of six?
The privacy workflow logs who triggered AI analysis, what information went to which provider and when. Enterprise administrators can export document activity and enable prompt history exports, while Word preserves accepted edits and comments. These records identify parts of the assisted work performed on an agreement. ContractKen also markets shorter drafting and review times, but gives no corresponding guidance on recording billable hours, charging clients or disclosing the effect of those savings.
The license expressly permits ThoughtRiver as part of managed services for a customer's legal clients. Marketing compares minutes of assisted review with hours of lawyer work, and the named Shoosmiths testimonial concerns complex supply agreements. Those savings can affect a client's bill as well as internal capacity. A record designed to allocate assisted work to billable time, or guidance on fees and client disclosure, is not described.
Outside Counsel Guideline Readiness
Can a firm get this vendor through a client’s AI clause without a bespoke negotiation?
ContractKen offers a DPA, current subprocessor register, security paper, data flow diagram and insurance certificates on request. Some assurance materials require a nondisclosure agreement. Its public security FAQ explains tenant separation, model provider restrictions and purge options, but is not a complete provider list or a client consent package. A firm can begin its assessment from those commitments, while the detailed materials for a client's AI requirements require direct contact.
The public register identifies Microsoft Azure in the United Kingdom, SendGrid in the United States, and Okta, ABBYY Vantage and Zoho Desk in Europe. It gives functions, data categories and transfer safeguards for those providers. The list supports part of a client's data handling review, but does not name the general purpose language model providers or supply a ready made client AI consent and disclosure pack.
Court Disclosure Support
If a judge’s standing order requires an AI disclosure, can the product produce one?
Anonymized versions and interaction logs preserve the information sent to AI providers, while drafting identifies source clauses and Word retains accepted edits. Administrators can export activity logs and prompt history. These provide pieces of an explanation of how an agreement was produced. A single export tying each generated passage to the exact model version, retrieved sources and a named human verification step for a court certification is not specified.
The Digital Issues List preserves flagged issues, risk levels and resolution status and can be shared or downloaded. That gives a contract reviewer a record of what required attention and how the review progressed. Word changes and collaboration add further context. An export combining the exact model, retrieved sources and human verification for each generated passage is not described, so these records do not by themselves form a complete AI disclosure certificate.
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.
- Ethical Walls and Matter Segregation
- Good Law Verification
Which one fits
Choose ContractKen if
- Your main workload combines drafting from source material with negotiation of the current Word agreement.
- Playbook fallbacks and walkaway positions should guide what returns to a lawyer or partner for judgment.
- You want to estimate a per user standard Review subscription and consider local masking or private inference separately.
Choose ThoughtRiver if
- A shared legal department deployment needs unlimited users, portfolio analysis and a package sized by review volume.
- You want a described classification evaluation process with separate measures rather than only a general speed claim.
- Connections to Microsoft 365, Power BI or legal project systems are relevant to the surrounding contract workflow.
In summary
ContractKen
ContractKen uses firm playbooks or a comprehensive clause library to review an agreement in Word, and adds drafting from precedents and term sheets. Its negotiating rules can include preferred positions, fallbacks, walkaway limits and escalation. The AI Legal Index identifies that combination of drafting inputs and explicit negotiation boundaries as the main difference from a broader portfolio analysis platform. Its standard Review plan charges by user and includes unlimited review and redlining. Local masking and restoration can protect sensitive terms before configured model processing, and private inference is an enterprise option. The default terms still reserve service improvement rights and do not establish a universal zero retention configuration.
ThoughtRiver
ThoughtRiver interprets agreements through Lexible, a legal model combined with other language models, then applies a team's playbook and proposes issues and redlines. Its platform connects to Word, Microsoft 365 and portfolio analysis tools. The AI Legal Index identifies its described evaluation process, with repeated tests against verified legal examples, as a useful distinction from an unsupported accuracy claim. Professional pricing starts at £15,000 annually in the United Kingdom for teams below 50 monthly reviews, with different displayed rates elsewhere. The platform has a private database instance and unlimited users. The agreement and DPA permit specified training uses, including anonymized contract text.
Questions buyers ask
Are ThoughtRiver's prices the same in every country?
Its pricing logic shows Professional from £15,000 and Enterprise from £30,000 per year in the United Kingdom. Outside the United Kingdom, the displayed starting rates are $20,000 and $35,000. These are the stated currency amounts, not conversions. From the AI Legal Index, based on each vendor's own published materials as of October 15, 2026. No vendor pays for placement.
What do ThoughtRiver's accuracy figures measure?
The current explanation gives F1, precision and recall for its legal classification process and describes repeated tests against verified examples. Those measures are distinct from the correctness of every generated redline or legal answer, and the full dataset is not available for independent reproduction. From the AI Legal Index, based on each vendor's own published materials as of October 15, 2026. No vendor pays for placement.
Does a private database mean contracts cannot be used for training?
No. ThoughtRiver's terms license specified training data and anonymized contract text, and its DPA names machine learning training. A private database concerns deployment and separation. ContractKen also distinguishes private inference from its separate service improvement permissions. From the AI Legal Index, based on each vendor's own published materials as of October 15, 2026. No vendor pays for placement.
Which charges for each reviewing user?
ContractKen's standard Review subscription is per user with unlimited review and redlining. ThoughtRiver's displayed packages include unlimited users and distinguish contract volume. Implementation and custom services can require separate terms, so the two starting figures do not price the same unit. From the AI Legal Index, based on each vendor's own published materials as of October 15, 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.
ThoughtRiver and ContractKen did not review this comparison. ThoughtRiver's country specific starting rates are stated in their original currencies. Neither vendor's classification or architecture claims establish guaranteed legal accuracy.