Dazychain vs Plexus: how they compare in 2026
Dazychain and Plexus both sell in house legal teams a front door for business requests, matter management and contract workflows, and both are built in Melbourne. Dazychain, made by Yarris Technologies, lists its prices, from $105 per user a month across three plans. Its AI runs only when a user clicks, inside the matter or document it was opened on, on Amazon Bedrock in Australia or the United States. Plexus quotes prices per Professional User license and puts more of its product into AI. The first version of each new contract gets an automatic review against the customer's precedents, and Plexus Counsel answers questions across up to ten documents under the legal team's playbooks. Marketing Wizard rates marketing assets against Australian advertising law. Plexus runs on OpenAI's enterprise models, with contract review on Amazon Bedrock in Australia, and its Subscription Terms carry a data processing addendum naming 19 subprocessors. Dazychain's AI runs on Bedrock with no model named, under terms it signs with each customer.
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.
Dazychain sells matter management first, described in its footer as capture, triage, manage, resolve. Its pricing table shows a complete platform on every plan, with AI as one section among matter management, documents, spend, collaboration, workflow, reporting and integrations. AI use is capped on Launch and unlimited on Scale and Optimize. DazychainAI runs only when a user clicks a function, and the AI FAQ says it can be turned off completely for an account, which leaves a working matter management system. The AI summarizes matters and documents, finds information across a matter's content, answers questions about the customer's documents and compares an external contract with the customer's playbook. Saved AI summaries and outcomes can be reused in the matter. All of it sits on top of the matter record rather than forming it.
AI runs through a contract, matter and approvals platform that also works without it. With AI features on, each new contract gets an automatic review when its first version is uploaded. The review holds an executive summary, a risk summary covering legal, commercial and operational risk, and a clause table. The table compares the draft with the customer's precedents, or with general drafting principles where none are loaded. Summaries sit in each document's Key Facts tab, and intake autofill reads an uploaded contract for party names, dates, values and jurisdictions. Marketing Wizard is where the model carries the work, rating each marketing asset high, medium or low risk against more than 230 Australian laws, codes and regulations. Its November 2025 launch said compliant content is approved automatically, and the product page says only high risk items reach legal. Intake forms, templates, approval routing, signing, the repository and Insights reporting all run with AI switched off. AI features arrived in 2023, six years after the platform launched in March 2017.
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.
Dazychain's AI FAQ says extractions are designed to be highly accurate but, like any automated system, may need review. Users validate and edit extractions in the platform before anything is finalized or shared. A limitations entry adds that the AI works only within the matter or document it is engaged on, and that clear, well structured and complete documents give better results. Dazychain gives no accuracy figure, test set or error rate, and does not describe how a summary, extraction or answer points back to the passage it came from. The AI page credits Amazon Bedrock with detecting correct responses and minimizing hallucinations and ambiguity. That is a statement about the hosting service rather than a measure of DazychainAI's own output on matters and contracts.
Grounding differs by feature. Contract review compares a draft with precedents the customer uploads, and its Confirmed Clauses list shows what matched the customer's standard. Since March 2026 Plexus Counsel shows the name of the Knowledge material behind an answer, and Marketing Wizard recommendations carry citation links to the Australian legislation behind them. Where no Knowledge is loaded, features fall back to the broader model, and document summaries and Draft Advice always use it. No accuracy measure accompanies contract review, summaries, redlines or Marketing Wizard. The one figure comes from Plexus's own testing of intake autofill, which it says fills about 75 percent of custom questions from the document. The AI Security Fact Sheet names hallucination and lists the mitigations: grounding in the customer's documents, prompt guardrails, tags on all AI generated content and review by a person. The Counsel help article warns that answers may include details that seem relevant but are not in the document.
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.
DazychainAI acts only when a user engages it. The AI FAQ says it accesses data only when a user actively engages an AI function, such as clicking a button, so confidential matters stay untouched until assistance is requested. It works only within the matter or document it is engaged on and does not reach data outside that context. Extractions may need review, and users validate and edit them in the platform before anything is finalized or shared. AI can be turned off entirely for an account, and the FAQ's answer on restricting AI by role or matter type offers only that account wide switch. Outputs carry no confidence indicator, and no threshold or abstention state is described. Separately, the intake feature lets business users generate a document such as a nondisclosure agreement from a template the legal team approved in advance.
Most AI output waits for a person. Redline suggestions arrive as cards showing the current wording, the proposed wording and the reason, stay visible only to the user, and apply as tracked changes once accepted. AI comments follow the same rule, and since August 2026 redlining asks the user to confirm details it is unsure of, such as an address, rather than filling them in. Draft Advice, offered to select organizations, is generated when a matter is raised and stays visible only to the legal team until a lawyer posts it. A warning appears when it is posted unedited, and once posted it is emailed to the matter's stakeholders. AI filled intake answers are marked, and users are told to check each one because answers can trigger delegation of authority rules. Playbooks are reviewed by their creator before publishing and can route a question to a named person in legal. Two features act without a person. Contract review runs automatically on the first version of each new contract, and Marketing Wizard's launch materials describe compliant content being approved automatically, while its help article routes submissions to a legal stakeholder for approval. How a risk rating is set, and what happens to a medium risk item, is undescribed.
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.
Dazychain's client stories name the organization and the people involved. Helen Carr, Deputy General Counsel at MYOB, describes using Dazychain's matter data to show growth in the volume and complexity of matters and to win additional headcount. Mercia Chapman, Senior Legal Counsel at Equity Trustees, speaks on email capture and retrieval. Further stories cover Perenti's global mining legal work, SEE Civil's time saved finding files, and process improvements at Sage Automation, with VicReturn, Education Services Australia and SAGE Group also listed. The MYOB account describes an operational outcome that the platform's reporting made possible. The stories are not dated, and no quantity is attached to any customer's result.
Named customers report figures. ENGIE Australia's head of legal and general counsel describe a 90 percent cut in contract execution time, a 50 percent year on year rise in executed volume and workflows finishing in 4.5 hours on average. That covers more than 320 users across 13 business units. Caidya's chief legal officer describes growth from 125 users to about 1,500 across more than 40 regions. Since October 2023 the team has handled more than 15,000 documents and more than 700 legal requests, answered in under two days on average. Lion's marketing team reports campaign approvals up to 90 percent faster and trade promotion permit costs up to 50 percent lower, across about 1,000 promotions since 2017. Shorter customer cards cite Noumi, with NDAs executed in 1.5 hours, LocalEyes, with promotion approvals seven times faster, and PMG, with approved terms in 24 hours. The case studies carry no dates or method, and Caidya's region count reads 41 in its case study and 43 on the customers page. Plexus says more than 200 legal teams use it, and its about page says more than 250 companies.
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.
Dazychain's security page describes data segregation as fully supported and tenant aware, so no customer can reach another's information unless permission is given, as in a collaboration. Data is encrypted with AES 256 at rest in a MongoDB store and with HTTPS and TLS 1.3 in transit, and documents sit encrypted in Amazon S3. Role based access applies on every plan, and every access and change to documents, matters and deliverables is logged and preserved. DazychainAI reaches nothing until a user clicks and stays inside the matter or document engaged. Amazon Bedrock does not store its inputs or outputs, each model runs in a dedicated AWS account, and customer information is not used to train external models. These commitments sit in the FAQ and the security page rather than a contract on the site. Staff sign confidentiality agreements and take annual security training, and privilege and work product get no separate treatment.
Customer data stays out of shared models by policy. The security page, the AI page and the AI Security Fact Sheet say customer data never trains public or shared AI models. The Counsel help article says it never trains or improves the model for anyone else. Each customer sits in its own databases and vector stores, which Plexus describes as zero data crossover between tenants. Staff reach customer data only with the customer's approval or under authorization from senior management and security. The Subscription Terms bind both parties to hold each other's confidential information in trust and use it only for the agreement, and their data processing addendum limits processing of personal data to the customer's instructions. Neither contract mentions AI training. Inside an account, permission sets grant review and view rights per document type and organization unit, and legal teams can leave comments on matters that only legal can see. The Draft Advice article says OpenAI did not keep the data or use it for training. Privilege and work product go unmentioned.
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.
Dazychain draws no advice line on its site, although DazychainAI drafts summaries and answers questions about legal documents. The only legal document the site links is a privacy policy, with no terms of service, end user agreement or disclaimer page. The AI FAQ notes that extractions may need review and that users validate them before anything is shared. It does not say that output is not legal advice or that a lawyer must review generated content before it is relied on. Business users who are not lawyers can complete an intake form that feeds a template approved in advance and generates a tailored document, such as a nondisclosure agreement. The AI page says that document can be reviewed or sent back to the requester without manual intervention, and sets no legal checkpoint around that choice. No rule of professional conduct, bar authority or jurisdiction limit is named.
The advice line is drawn in the contract and in the product. The Subscription Terms state that the platform is not intended to constitute, or be a substitute for, legal advice, and make the customer solely responsible for the documents and results it generates. The AI Security Fact Sheet calls output general guidance rather than final legal advice and says anything with legal or commercial consequences should be reviewed by someone competent to judge it. Marketing Wizard results say they are not legal advice and do not replace a full legal review, and the tool is intended for the Australian market only. Draft Advice warns that AI may produce incorrect, non compliant, non standard or incomplete content. Plexus also practices law. Under its Subscription Terms Promotion Wizard counts as legal advice, covered by Plexus's professional insurance as a law firm. Marketing Wizard work from customers without an in house legal team goes to the Plexus legal team. Plexus Counsel is marketed as a digital lawyer. How the product supports a customer lawyer's competence and supervision duties is unaddressed.
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.
Dazychain's AI FAQ covers what DazychainAI does, where it runs, what data it uses, how to switch it off and its limits, and its security page covers information security. Neither sets out AI principles, an owner accountable for model behavior, testing before release or findings on bias. The nearest structure is the Information Security Management System, with documented policies, annual staff training and documented acceptance, which governs staff conduct and data security rather than model behavior. DazychainAI's limits run through product design instead. The AI stays inactive until a user clicks, works inside one matter or document and can be switched off for the whole account, and the FAQ offers no narrower control by role or matter type. Dazychain holds no ISO/IEC 42001 certification.
No AI governance framework sits behind the product. There is no responsible AI page, no named owner for model behavior, no description of testing before release, no results on output quality and no ISO/IEC 42001 certification. The AI Security Fact Sheet says AI supports rather than replaces human decisions and that all AI generated content is tagged, and the AI page says every AI output is auditable. The rest of the security material covers ISO 27001, SOC 2, encryption and tenant isolation, which govern information security rather than model behavior. Whether review quality differs by contract type, jurisdiction or language is unaddressed. Marketing Wizard is built only for Australian law, while Plexus serves enterprises across Australia, New Zealand and Singapore alongside global consumer brands.
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.
On termination Dazychain keeps customer data for the period the customer requests, usually one month, then deletes it, with other arrangements on request. The FAQ gives backup retention of two weeks for the MongoDB database and unlimited retention for documents in Amazon S3. Access control runs through role based access on every plan and an audit trail that logs every access and every change to documents, matters and deliverables. Staff pass police background screening on joining and every two years, sign confidentiality agreements and take annual security training. Data is encrypted with AES 256 at rest and TLS 1.3 in transit. Amazon hosts the application and files and Object Rocket the database. The FAQ says all subprocessors, such as AWS, ObjectRocket, CloudFront and SIEM providers, are bound by contractual controls, without listing them all. Penetration testing runs internally before each release and annually through an outside agency, and AWS multiple availability zones serve as the disaster recovery setup. Dazychain states compliance with the Australian Privacy Act 1988 and GDPR. Dazychain sets no breach notification period or commitment on its site.
Data handling is set out across the security pages and the Subscription Terms. Data is encrypted with TLS 1.2 or later in transit and AES 256 at rest, encrypted backups allow recovery to any point in the previous seven days, and uploads are scanned for viruses and malware. Independent security firms run penetration tests, and staff reach customer data only with the customer's approval or under authorization from senior management and security. The data processing addendum in the Subscription Terms commits each party to report a personal data breach without undue delay. It deletes personal data at the end of the agreement and promises notice before a new subprocessor is engaged. Its Annex III names 19 subprocessors, including Amazon Web Services, Heroku, Google Cloud, DocuSign, Auth0 and OpenAI, without purposes or locations. Retention periods for documents, prompts and outputs are unstated, and files uploaded to a Counsel thread stay available for the life of the thread. The older data security page, last reviewed July 2023, still cites ISO 27001:2013, and the privacy policy carries no breach notice of its own.
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.
Dazychain's site links a privacy policy and a sitemap as its only legal documents, with no terms of service, subscription agreement, end user terms or service level document. The FAQ refers to data processing agreements and a 12 month minimum contract term, which sit in documents signed with each customer. Indemnities, liability caps, warranties, exclusions and insurance are left to that signed contract. The AI FAQ describes output as designed to be highly accurate while noting that it may need review. How loss from a wrong summary, extraction or playbook comparison is shared between Dazychain and a customer depends on the contract offered in a sales process. The privacy policy is the only document on the site that governs customer data.
Two documents set the terms, and both cap and disclaim. The Subscription Terms, which Plexus says it does not amend, cap total liability at the fees paid in the 12 months before a claim. They exclude lost profits, lost data and indirect loss, exclude implied warranties, and make the customer solely responsible for documents and results the platform generates. The Legal Gateway Terms of Use, which carry no date, set the same 12 month cap with remedies at Plexus's election and leave reliance on generated material at the user's own risk. Their warranty exclusion is subject to warranties implied by the Competition and Consumer Act 2010. Neither document carries an indemnity from Plexus. The AI Security Fact Sheet points to an intellectual property indemnity that OpenAI gives under its enterprise terms. Promotion Wizard output sits under Plexus's professional insurance as a law firm, while the software carries no insurance or service level commitment. Victorian law governs for customers in Australia, New Zealand and Asia Pacific.
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.
Dazychain's integrations page and FAQ name its connections and what each does. Outlook creates matters from emails and files messages into existing matters, including from shared mailboxes, and Gmail does the same without shared inbox support. Word documents are created, edited and redlined inside matters with version history, and DocuSign sends documents for signature from a workflow. The FAQ adds ServiceNow and single sign on through Okta, Microsoft Entra and other SAML providers, and the pricing table lists SCIM and ADFS. The pricing table shows API access on Optimize, while the FAQ says API and webhook capabilities are not available, and the integrations page offers API based HR system integrations on request. SharePoint and a document library and knowledge base are listed as coming soon. The pages describe what each email and document connection does, without configuration detail.
Word and Outlook are the main connections. The Word add in, released in December 2025, creates documents from templates, saves them as contract records and uploads new versions. Since July 2026 it also runs AI review and redlining inside Word for users with a Microsoft 365 business license. The Outlook add in, also released in December 2025, turns emails into contracts or matters, attaches conversations and sends documents for signature, keeping documents and contract data in sync. Plexus announced both on Microsoft's marketplace in February 2026. Word Online editing has run since 2023, and a Gmail add in raises matters. Signing runs through an embedded DocuSign integration or a customer's own DocuSign Part 11 account for organizations under FDA rules. Single sign on works with Entra ID, ADFS, Okta, Google and SAML, across up to 1,000 domains. There is no public API yet. Plexus's roadmap, dated July 2026, targeted a documented platform API for September 2026 and a Legal MCP server for Claude, Copilot or ChatGPT for October 2026, and the release notes through September 2026 show neither.
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.
Dazychain's security page places application data in Amazon data centers in Australia and the United States, the database with Object Rocket in the same two regions, and documents in Amazon S3, across multiple availability zones. The AI runs on Amazon Bedrock in Australia or the United States according to the customer's location, with each model in a dedicated AWS account, so processing is addressed apart from storage. Older FAQ entries and the privacy policy say data is stored in Australia, and the privacy policy says personal information is not disclosed outside Australia without consent. Segregation is described as tenant aware, without describing the architecture. No single tenant, dedicated or self hosted option is offered, no region choice is presented to the buyer, and deployment does not change between the three plans. Dazychain has offices in Melbourne and Boston.
Plexus runs as a multi tenant service on AWS, with each customer in its own databases and vector stores. Personal information is stored on servers in Australia, and the privacy policy says data may also sit in the United Kingdom and the United States. Since August 2026 AI contract review runs on Amazon Bedrock in Australia, which Plexus ties to data residency and sovereignty requirements. The Counsel launch in November 2025 described Australian infrastructure with residency controls for customers that need them. The other AI features call OpenAI's enterprise API, and where that processing happens is unstated. Region choice is not a product setting, and there is no single tenant or private deployment. The contract management page gives implementation times of one to two weeks, and two to three weeks on average to full deployment.
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.
Dazychain states that its ISO 27001 controls and wider Information Security Management System are audited internally and externally each year. It also states that SOC 2 internal and external audits run annually and produce SOC 2 reports. No auditor or certifying body is named, and no SOC 2 type is given, so the page does not say whether the report covers a point in time or a period. The certificate and reports carry no date, period or scope, and Dazychain offers no trust portal or request form for them. The ISO claim cites the 2013 edition of ISO 27001, whose transition period to the 2022 edition ended in October 2025. AWS certifications listed alongside, including FedRAMP and SOC 1, belong to AWS. Penetration testing runs internally before each release and annually through an outside agency.
Plexus states ISO 27001:2022 certification on its security page and in its information security policy, last reviewed March 2024. Its AI page and AI Security Fact Sheet add SOC 2 Type 2 compliance, which the security page leaves out. No auditor, certificate number, scope or date appears with either. A security portal holds the disaster recovery report, the ISO 27001 certificate, a penetration test attestation letter and network security, risk management and secure development policies. Access comes through a request form, and verification takes up to three business days. The portal's document list has no SOC 2 report. Penetration tests come from independent security firms Plexus does not name. The older data security page, last reviewed July 2023, cites the 2013 edition of ISO 27001.
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.
All DazychainAI tools run on Amazon Bedrock, with each model in a dedicated AWS account. Bedrock does not store inputs or outputs, share them with third parties or use them for training, and inference runs in Australia or the United States according to the customer's location. The service, the region and the data conditions are stated, while the model maker is not. Bedrock hosts models from several makers, and Dazychain names no model or version and does not say which model produces a summary as against a playbook comparison. Dazychain sets no notice period for a change of model, version or hosting arrangement, and its FAQ binds subprocessors such as AWS through contractual controls without listing them.
OpenAI is the named provider. The security page, the AI page and the AI Security Fact Sheet describe OpenAI's enterprise GPT models behind Plexus AI, and the Subscription Terms list OpenAI among the subprocessors. Since August 2026 AI contract review runs on Amazon Bedrock in Australia, with the model there unnamed. The Draft Advice article, dated August 2026, still refers to OpenAI's GPT 3 API, and Marketing Wizard's launch post describes a proprietary AI engine. No model version is mapped to a feature. A change of provider carries notice, since the data processing addendum promises notice before a new subprocessor and its transfer clauses set one month's warning. A change of model within a provider carries none. The Draft Advice article says data sent to OpenAI was not kept or used for training.
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.
Dazychain's pricing page lists three plans per user per month, Launch at $105, Scale at $130 and Optimize at $160. A currency selector offers Australian dollars, United States dollars, euros and pounds, and the figures carry a bare dollar sign. Launch carries no implementation fee. Scale offers customized configuration and implementation as a $5,000 option, and Optimize lists it at $5,000, while the FAQ says implementation is completely free. Reconfiguration is a professional services fee on the lower plans, and Optimize includes one each year. A feature table of about forty rows splits matter management, documents, spend, collaboration, AI, workflow, reporting, integrations, security and support by plan, and AI use is capped on Launch and unlimited above it. A 14 day trial starts from a sign up form, followed by workspace setup and an hour of training. The FAQ sets a 12 month minimum term and says Dazychain does not usually offer pilots.
Prices are quoted on request. The help center defines the unit of charge as the Professional User license, charged per person who meets the definition. The Subscription Terms set fees by the number of Professional Users and Workflows in the order form and bill overage pro rata. Fees rise by 7 percent plus CPI on each anniversary, and contracts renew automatically for 12 month periods. A free trial with no payment card includes Approve and eSign, document storage, legal requests, Marketing Wizard and Plexus Counsel, with no stated length. There are no tiers, feature splits or implementation fees on the open pages. Promotion compliance is sold at 50 percent below traditional legal review with a guaranteed 24 hour turnaround. Random Draw, the trade promotion draw service, asks for sign in only at payment and caps prizes at $200,000.
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.
Dazychain is built for corporate legal departments, and its FAQ says it is designed for small to medium sized in house legal teams. Pricing per user lets a team size itself. Its case studies name MYOB, Equity Trustees, VicReturn, Education Services Australia, SAGE Group, Sage Automation, Perenti and SEE Civil, mostly Australian organizations. The same platform is sold to HR teams under a separate product line, with case types such as investigations, performance management and protected disclosures, and Dazychain describes a front door spanning legal, HR and finance. The system works with any matter type, and the FAQ names charities, non profits, research and utility companies, superannuation and finance, construction and mining among the industries it serves. Dazychain has offices in Melbourne and Boston and hosts data in Australia and the United States. No jurisdictional coverage, team size limit or work the product is not suited to is stated.
Plexus serves in house legal teams in larger companies, with marketing, sales, procurement and operations staff using it to request legal help, approve marketing and run promotions. Its case studies come from energy, consumer goods, clinical research and lending, and its pages show consumer brands such as Nike, Woolworths and Nestlé. Plexus describes its enterprise base as spanning Australia, New Zealand and Singapore. The Subscription Terms contract through Plexus Gateway for Australia, New Zealand and Asia Pacific and through Legal Gateway International for the rest of the world. Marketing Wizard covers Australian advertising law only, and customers without an in house legal team can send marketing reviews to Plexus's own lawyers. Law firms are not a target segment, and no team size or contract type is ruled out.
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?
Dazychain's AI FAQ says customer information is never used to train external AI models, and that the platform uses only the data within the document or matter being worked on, combined with Amazon Bedrock's services. The AI page adds that Bedrock does not use inputs or outputs for training. Each statement covers external models, and Dazychain says nothing about training, tuning or evaluating models of its own or of its parent, Yarris Technologies.
The site links no terms of service, subscription agreement or end user terms, so these commitments sit in the FAQ and the AI page rather than in a contract term. The privacy policy and a sitemap are the only documents linked from its navigation and footer. A customer can switch DazychainAI off for its whole account, so no matter data reaches the models.
Customer data stays out of public and shared models by policy. The security page and AI Security Fact Sheet say customer data never trains public AI models, and the AI page says it never trains public or shared models. The Counsel help article says it never trains or improves the model for anyone else. The Subscription Terms and the Terms of Use leave training unmentioned. They bind Plexus to use confidential information only for the agreement and to process personal data on the customer's instructions, which limits use without naming AI.
The privacy policy mentions integrated AI tools for document discovery and scanning without a training position. Whether Plexus trains models of its own on customer content is unstated. The Draft Advice article says data sent to OpenAI was not used to train its models.
Prompt and Output Retention
How long does the product keep what a lawyer typed, and can that be set to zero?
Amazon Bedrock does not store DazychainAI inputs or outputs, according to Dazychain's AI page, and each model runs in a dedicated AWS account. Saved AI summaries and outcomes become part of the matter record and can be reused. On termination, customer data is kept for the period the customer requests, usually one month, then deleted, with other arrangements available on request. The FAQ gives backup retention of two weeks for the database and unlimited retention for documents in Amazon S3, where documents are encrypted with AES 256.
Dazychain sets no retention period during the subscription, on any plan, for matter records or the AI output saved into them. The audit trail keeps every access and change to documents, matters and deliverables.
Retention is acknowledged without a period. Files uploaded to a Counsel thread stay available for the life of the thread and do not join the document list, and accepted AI summaries, comments and redlines become part of the document record. The Draft Advice article says OpenAI did not keep data sent through its API. The data processing addendum deletes personal data at the end of the agreement unless European or UK law requires it to be kept.
Encrypted backups allow recovery to any point in the previous seven days. There is no stated period for prompts, Counsel conversations or outputs during the subscription, and no customer setting controls it.
Ethical Walls and Matter Segregation
Does retrieval respect the firm’s ethical walls, or can the model read across them?
The security page states that data segregation is fully supported and tenant aware, and that no user company can access another's information unless permission has been provided, as in a collaboration. Role based access applies, with every access and modification logged. On the AI side the model reaches nothing until a user engages a function, and it works only within the matter or document engaged rather than across the account.
There is no wall at matter level inside a single legal department. Dazychain describes no conflicts management, restricted matters or walled user view. The vendor's FAQ asks whether AI can be restricted by role or matter type and answers only that it can be switched off for the whole account.
Plexus runs its own permission model. Core, full and custom permission sets grant review and view rights per document type, scoped to the company's structure, and since May 2026 legal teams can leave comments on matters that only legal can see. Insights can hide sensitive document types, and only users with account administration rights can upload or change the Knowledge the AI draws on. Draft Advice stays visible to the legal team until a lawyer posts it.
No permission is specific to AI features. Between customers, each tenant has its own databases and vector stores, and Plexus staff reach customer data only with the customer's approval or under authorization from senior management and security.
Third Party Request and Subpoena Notice
If someone subpoenas the vendor for a firm’s data, does the firm hear about it first?
Dazychain's privacy policy says it will not disclose personal data to government institutions and authorities except where the law requires, and that it will not disclose personal information outside Australia without consent. It states no commitment to tell the customer when a disclosure is required. Any contractual notice term would sit in the data processing agreements the FAQ says Dazychain signs with customers, and Dazychain issues no transparency report.
The FAQ adds that subprocessors such as AWS and ObjectRocket are bound by contractual controls. The security page says data is not shared with third parties and that users may choose to share information with external lawyers. Hosting spans Australia and the United States on the security page, so matter data can sit under two disclosure regimes, while the privacy policy says Dazychain stores its data on servers in Australia.
The Subscription Terms commit to notice. Clause 6.4 allows confidential information to be disclosed where law, a regulator or a court requires it, provided Plexus gives as much notice as possible where legally permitted. It also takes the customer's reasonable requests on the disclosure into account. The standard contractual clauses attached to the data processing addendum add prompt notice of a public authority request for transferred personal data, a review of its legality and periodic information on requests received.
The privacy policy, effective May 2023, separately allows disclosure to regulators, government authorities and parties handling a criminal investigation. There is no transparency report. Personal information is stored in Australia and may also be held in the United Kingdom and the United States.
Primary Law Corpus Provenance
Where does the law in this product come from, and does the vendor have the right to use it?
DazychainAI works only on the customer's own material. The AI FAQ says it uses only the data within the document or matter being worked on. The features operate on uploaded contracts, matter records, emails and the customer's own contract playbook. The product retrieves no primary law, published precedent, market clause bank or licensed dataset, and names none. The playbook comparison is the one feature that measures a contract against a standard, and that standard is the customer's own playbook rather than content Dazychain supplies.
Intake document generation likewise fills templates the customer's legal team approved in advance, so no outside licensing question arises.
Most of the AI works from the customer's own material. Contract review compares documents with the customer's precedents, Counsel and Playbooks draw on Knowledge that administrators load as DOCX or PDF files under 25MB, and intake autofill reads the uploaded document. Counsel also draws on what its help article calls publicly available information and licensed data, without naming the sources or licenses. Marketing Wizard checks assets against more than 230 laws, codes and regulations for the Australian market and links each recommendation to the legislation behind it.
Which instruments make up that set, and how they are kept current, is unstated. Where no Knowledge is loaded, features use the general OpenAI model, whose training data is undescribed.
Good Law Verification
Does the product tell you when the authority it just cited has been overruled?
Dazychain manages matters, documents and spend, and its AI summarizes, finds information in and answers questions about the customer's own documents and compares a contract with the customer's playbook. No case, statute or regulation appears to a user anywhere in the workflow, so a citator sits outside the product. The playbook comparison flags where an external contract departs from the customer's playbook, and keeping the playbook current is the legal team's work.
Plexus reviews and drafts contracts and marketing material and does not cite case law, so a citator sits outside the product. Marketing Wizard links its recommendations to Australian legislation, and how amendments to those laws reach its rules, or which version applied to an asset, is undescribed. When a precedent in Knowledge changes, users regenerate contract review recommendations by hand, and each later version of a contract needs its review regenerated the same way.
Refusal and Uncertainty Behavior
What does the product do when the answer is not in the corpus?
The AI FAQ's limitations entry says the AI works only within the matter or document it is engaged on and does not reach data outside that context. It adds that clear, well structured and complete documents give better results. A second entry says extractions are designed to be highly accurate but may require review, and that users validate and edit them before anything is finalized or shared. Neither entry says what the system does when it cannot ground an output, such as a scanned contract or a playbook point with no matching clause.
Outputs carry no confidence indicator, and no state for declining to answer is described. The FAQ asks whether AI can be restricted by role or matter type and answers only that it can be switched off for the whole account. DazychainAI acts only when a user engages it, so every output follows a user's request.
Two features hold back when the input is unclear. Intake autofill leaves a field blank when the contract does not contain the answer. Since August 2026 AI redlining asks the user to confirm details it is unsure of, such as an address, rather than filling them in. AI answers in intake are marked, and users are told to check each one because answers can trigger delegation of authority rules. Contract review does not run on documents with unresolved tracked changes, bulk uploads or legacy files, and reviews only the main document.
Plexus Counsel has no described way to decline a question, and its help article warns that answers may include details that seem relevant but are not in the document. Draft Advice warns a lawyer who posts it unedited.
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 Dazychain or its parent Yarris. DazychainAI summarizes and answers questions about the customer's own matter documents rather than producing court filings.
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 Plexus, Plexus Counsel or Legal Gateway. Plexus reviews and drafts contracts and marketing material for in house teams rather than court filings, and Marketing Wizard's citations point to Australian legislation.
Bar Guidance Alignment
Has the vendor engaged in public with the ethics opinions its buyers are bound by?
Dazychain names no bar authority, conduct rule or ethics opinion, and maps no guidance to a jurisdiction. Its compliance material covers data protection, running to the Australian Privacy Act 1988 and related state laws, GDPR, ISO 27001 and SOC 2. One feature touches professional responsibility directly. Business users who are not lawyers can complete an intake form that generates a finished document, such as a nondisclosure agreement, from a template approved in advance.
Dazychain says nothing about the legal department's duties when it approves such a template or lets the document go out without review. DazychainAI's limits, such as acting only when a user clicks and only inside one matter, are presented as security and data controls rather than tied to a lawyer's duty of competence or supervision.
The product's language on legal advice is general. The Subscription Terms state that the platform is not a substitute for legal advice, and the AI Security Fact Sheet calls output general guidance rather than final legal advice. It asks that anything with legal or commercial consequences be reviewed by someone competent to judge it, and Marketing Wizard results say they are not legal advice. Promotion Wizard is legal advice from Plexus's own lawyers under its professional insurance as a law firm. No bar opinion, ethics rule or professional conduct guidance is named anywhere in the product material.
Billing and Fee Posture
Does the vendor address what happens to the bill when the work takes an hour instead of six?
Dazychain carries spend and budget management, invoice review workflows, spend analytics, schedules of rates and budget tracking for what the department pays its external law firms. The AI material says the platform helps lawyers save time, removes friction from drafting, reporting and summarization, and lets teams get work done in a fraction of the time, with no figure attached. Dazychain draws no link between that time saved and what outside firms bill.
The matter record does not mark which summaries or playbook comparisons the model produced, either in the matter or in reports to the business. Invoice review runs as a workflow beside the department's schedules of rates and budgets.
Plexus is sold to in house legal teams and the business teams they support, so no client bill sits in the loop for its main buyers. Its savings claims address the buyer's own costs: a 65 percent lift in legal productivity, 90 percent faster contract execution and 70 percent fewer routine tasks on the home page. For marketing compliance it claims 30 percent lower external counsel spend and up to 70 percent fewer routine escalations, none with a stated method.
Promotion compliance, reviewed by Plexus lawyers, is priced at 50 percent below traditional legal review with a 24 hour turnaround. Fee treatment of AI assisted work for a firm that bills a client is outside the product's scope.
Outside Counsel Guideline Readiness
Can a firm get this vendor through a client’s AI clause without a bespoke negotiation?
Dazychain's security and AI pages place the application and files with Amazon in Australia and the United States, the database with Object Rocket and AI on Amazon Bedrock. Each model runs in a dedicated AWS account with no storage of inputs or outputs. The FAQ says all subprocessors, such as AWS, ObjectRocket, CloudFront and SIEM providers, are bound by contractual controls, and refers to data processing agreements signed with each customer.
The site carries no subprocessor list or agreement. Bedrock hosts models from several makers and Dazychain names none, so the model that reads a contract is unidentified. Dazychain offers no summary written for a legal department to pass to its business, auditors or regulators.
The documents a counterparty's AI clause asks for sit in the Subscription Terms, readable without an agreement. Annex III of the data processing addendum names 19 subprocessors, including Amazon Web Services, Heroku, Google Cloud, DocuSign, Auth0, OpenAI, InfoTrack, Zoho and Intercom, without purposes or locations. The addendum, with standard contractual clauses attached, promises notice before a new subprocessor is engaged.
Cloudflare, named on the data security page for web protection, is not on the list, and the AWS entry covers the Bedrock hosting behind contract review. Further security documents, including a penetration test attestation letter and the ISO 27001 certificate, sit in a portal reached through a request form.
Court Disclosure Support
If a judge’s standing order requires an AI disclosure, can the product produce one?
The security page says every access to the platform and every modification to legal documents, matters and deliverables is logged and preserved, as a safeguard against unauthorized activity. Audit trails and activity tracking appear as an included feature on all three plans, alongside exportable reports and dashboards, so an exportable record of what happened to a matter exists. The log does not mark which summaries, extractions or playbook comparisons came from the model rather than a person.
Saved AI summaries and outcomes stay in the matter record, where each later change is logged. No model or version is attached to any output, and Dazychain offers no guidance or template for disclosing AI use to a court, regulator or client.
Plexus says every AI action is tracked and reviewable, and all AI generated content is tagged in the document. The activity feed records Plexus AI summaries, documents keep a version history with side by side comparison, redlines apply as tracked changes once accepted, and Counsel names the Knowledge behind an answer. Document and matter lists export to CSV up to 10,000 rows at a time, and the Usage and Risk and Suppliers reports in Insights export to Excel.
There is no export of the AI activity log, record of the model used or disclosure template. Plexus's output is contracts and marketing reviews 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.
- AI Governance and Bias Disclosure
- Good Law Verification
Which one fits
Choose Dazychain if
- You want the price before the first call. Dazychain lists Launch at $105, Scale at $130 and Optimize at $160 per user a month, with a currency selector for Australian and United States dollars, euros and pounds. Launch carries no implementation fee, the upper plans list configuration and implementation at $5,000, and a 14 day trial includes an hour of training.
- You want AI that does nothing until someone asks. DazychainAI runs only when a user clicks a function and reads only the matter or document it was opened on. It can be switched off for a whole account, leaving intake, workflows, documents, spend and reporting running. Usage is capped on Launch and unlimited on Scale and Optimize.
- You want AI processing and storage in Australia or the United States. Dazychain hosts its application with Amazon, its database with Object Rocket and its documents in Amazon S3 in those two regions. Each AI model runs in a dedicated AWS account on Amazon Bedrock, which Dazychain says stores no inputs or outputs.
Choose Plexus if
- You want AI working on every new contract without being asked. Plexus reviews the first version of each new contract against the customer's precedents, flags missing or risky clauses and summarizes key terms. It fills intake fields from uploaded documents and offers redline cards in Word that change the document only when accepted, as tracked changes.
- Your legal team also clears marketing and promotions. Marketing Wizard rates each asset high, medium or low risk against more than 230 Australian laws and codes and sends high risk items to legal. Promotion Wizard produces trade promotion terms that Plexus's own lawyers review as legal advice, under its professional insurance as a law firm.
- You want the contract terms and subprocessors in writing before you sign. Plexus's Subscription Terms carry a data processing addendum with breach notice without undue delay, notice before a new subprocessor and an annex naming 19 subprocessors, including OpenAI. They cap liability at the fees paid in the prior 12 months.
In summary
Dazychain
Dazychain is matter management software for small to medium in house legal teams, made by Yarris Technologies of Melbourne. Requests arrive through intake forms, Outlook or Gmail and become matters with workflows, tasks, due dates and approvals. Around them sit document and contract management with Word editing, client and law firm portals, spend tracking and dashboards. DazychainAI summarizes, answers questions and compares contracts with a playbook when a user engages it. Three plans run from $105 to $160 per user a month, AI runs on Amazon Bedrock in Australia or the United States, and Dazychain states ISO 27001 and SOC 2.
Plexus
Plexus is a legal operations platform for in house legal teams, covering guided intake, contract drafting, approvals and signature through DocuSign, and marketing and promotion compliance. Plexus AI reviews the first version of each new contract against the customer's precedents, fills intake fields from uploaded documents and suggests redlines in Word that apply only when accepted. Plexus Counsel answers questions under the legal team's playbooks. The AI runs on OpenAI's enterprise models, with contract review on Amazon Bedrock in Australia. Prices are quoted per Professional User license, and ENGIE Australia, Caidya and Lion report figures in case studies.
Questions buyers ask
Dazychain vs Plexus: which is better for an in house legal team?
Dazychain suits a smaller team that wants listed prices, AI that runs only when asked and a matter system that keeps working with AI switched off. Spend tracking and a portal for outside firms are built in. Plexus suits a team that wants AI across contract review, intake and marketing compliance, contracts approved and signed in one flow, and published terms with a named subprocessor list, at a quoted price. Both are built in Melbourne for in house legal departments and charge per user. From the AI Legal Index, based on each vendor's own published materials as of October 9, 2026. No vendor pays for placement.
Do Dazychain and Plexus train AI on customer data?
Both say they do not train external or shared models. Dazychain's AI FAQ says customer information is never used to train external AI models and that Amazon Bedrock does not use inputs or outputs for training. Plexus says customer data never trains public or shared models and keeps each customer in its own databases and vector stores. Its Draft Advice article says OpenAI did not keep or train on the data. Neither statement sits in a customer agreement, and neither says whether the vendor trains models of its own. From the AI Legal Index, based on each vendor's own published materials as of October 9, 2026. No vendor pays for placement.
Where does each one run its AI?
Dazychain runs all AI on Amazon Bedrock in Australia or the United States, depending on where the customer is, with each model in a dedicated AWS account and no model named. Plexus runs AI contract review on Amazon Bedrock in Australia and its other AI features on OpenAI's enterprise models, without saying where that processing happens. Plexus stores personal information in Australia and says data may also be held in the United Kingdom and the United States. From the AI Legal Index, based on each vendor's own published materials as of October 9, 2026. No vendor pays for placement.
How much do Dazychain and Plexus cost?
Dazychain charges $105, $130 and $160 per user a month across three plans, with AI capped on Launch and unlimited above it. Launch has no implementation fee, the upper plans list $5,000 for configuration and implementation, the trial runs 14 days and the minimum term is 12 months. Plexus quotes prices per Professional User license and raises fees by 7 percent plus CPI each year. Its free trial needs no payment card and includes Plexus Counsel and Marketing Wizard. From the AI Legal Index, based on each vendor's own published materials as of October 9, 2026. No vendor pays for placement.
What do Dazychain and Plexus both leave unpublished?
Neither publishes an accuracy rate for its summaries, reviews or redlines, an AI governance framework or bar guidance on AI, and neither names the model version behind each feature. Neither offers an indemnity covering AI output or a customer choice of data region, and neither issues a transparency report on government requests. Dazychain publishes no customer agreement or subprocessor list, while Plexus publishes both in its Subscription Terms. From the AI Legal Index, based on each vendor's own published materials as of October 9, 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.
Plexus's training statements sit on its security page, AI page, AI Security Fact Sheet and help articles and cover public or shared models. Its Subscription Terms limit use of confidential information to the agreement without naming AI training. Dazychain's training statement sits in its AI FAQ and covers external models, and its customer agreement is signed directly rather than published. Plexus lawyers review trade promotion terms as a service alongside the software. Neither vendor reviewed this page.