Athennian vs DiliTrust: how they compare in 2026
Athennian and DiliTrust both sell entity and governance software to in house legal teams. Athennian is a dedicated entity management platform, while DiliTrust sells a five module suite that adds a board portal, contracts, matters and a dataroom. Both document their products in depth. The difference is where each puts its commitments. Athennian puts them in published terms: its AI terms bar it and its named model providers, Google and Amazon, from training any model on customer data. DiliTrust puts its commitments in an AI code of conduct and in certificates anyone can download: ISO 27001 and ISO 27701, both dated 23 February 2026. It states that its models are its own, fine tuned from open source models on synthetic and public data, and run in data centers in the client's region. It also names customers such as BNP Paribas and LVMH, where Athennian names none.
Which one fits
Choose Athennian if
- You want the training bar in the contract, covering the model providers too. Athennian's AI terms, part of its services agreement, state that neither Athennian nor its model providers may use customer data to train any model, and name Google Gemini and Amazon Bedrock among those providers.
- Your entity data must flow to other systems. Athennian offers a two way API with public developer documentation and developer terms, alongside Microsoft 365, electronic signature and document management integrations.
- You manage entities across many countries for legal, tax and treasury teams. Athennian states more than 500,000 entities managed across 150 countries, publishes global compliance guides, and its AI extracts governance data from uploaded documents and flags anomalies and missing data for review.
Choose DiliTrust if
- You want board, contract, entity and matter work in one suite. DiliTrust combines a board portal, contract management, entity management, matter management and a dataroom, with its Lini AI drafting board minutes, transcribing meetings, extracting contract terms and flagging risky clauses.
- Your procurement team wants certificates without a sales call. DiliTrust publishes its ISO 27001 and ISO 27701 certificates, dated 23 February 2026, and a Spanish ENS certificate as downloads, with a security sheet and its AI code of conduct, and offers its SOC 2 Type 2 report on request.
- Data sovereignty matters to your board. DiliTrust states that its data centers sit in the regions where clients operate, that its AI runs on its own infrastructure without third party sharing of personal data, and that organizations can connect their own language models through its API.
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.
Athennian combines entity and people records, ownership structures, appointments, minute books and compliance tasks with AI assisted governance work. Its agents extract information from source files and prepare changes to registrations, authority and entity profiles for approval. Intelligence answers questions across corporate material, with citations and optional restriction to selected entities. Read MCP extends current records into external AI assistants for ownership analysis, diligence and reporting. These capabilities reduce manual administration within a larger platform that still supports charts, templates, permissions and recordkeeping without generation. AI Intelligence and AI Core are separately priced additions, and the general workflow automation feature remains labeled beta.
Lini supplies AI functions across DiliTrust's five governance modules: contracts, entities, matters, board meetings and secure file sharing. Risk Detector applies internal playbooks and proposes tracked edits, while extraction fills contract summary sheets. Other functions transcribe meeting recordings, draft minutes, summarize files and generate QuickView matter updates. These functions extend systems that also manage permissions, calendars, agreements and corporate structures without generative processing. The AI is embedded in the suite rather than constituting the entire product. Customers can additionally connect their own language models through a secure API, subject to their chosen integration and governance arrangements.
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.
Intelligence retrieves information from minute book files and connects findings across entities. Each answer links to the relevant source section so a reviewer can inspect the underlying material. Users can restrict the source entities, and structured results can open in Custom Reports for filtering and further analysis. The system says it identifies missing information instead of guessing when the minute books lack an answer. Read MCP can also cite the live Athennian records used by an external assistant. No reproducible accuracy measurement, evaluation collection or distribution of failure types accompanies these claims, and corporate source links do not validate external legal authorities.
Risk Detector accepts a Word or PDF agreement, identifies its type and applies the customer's playbook rules. It highlights the clauses that depart from those rules alongside proposed edits, leaving the original contract language available for inspection before the reviewer accepts or adjusts a suggestion. This gives contract findings a concrete source and review path. The same assurance is not quantified across generated meeting minutes, general summaries or QuickView updates. No reproducible accuracy benchmark or error distribution is specified for those functions. A highlighted contractual passage also says nothing about the validity of an external legal proposition or current case authority.
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.
Entity agents extract proposed changes from source files but require a user to review and approve each update before it reaches the corporate record. Administrators can exclude entities from AI access, control intake visibility and set naming and filing preferences. Intelligence supplies direct source references for review. Read MCP has a separate categorical limit: connected assistants can retrieve authorized information but cannot add, change or delete Athennian data. Its setup assigns an API role and access groups, with returned findings reviewed in the connected assistant. The AI terms require human judgment before reliance and prohibit representing generated output as human work.
Users initiate AI work by supplying an agreement, asking a question or requesting a summary. Risk Detector returns highlighted clauses and tracked proposals that the reviewer can accept or adjust. Minutes Generation provides an editable draft assembled from meeting files and recordings; QuickView produces a matter update for the user to inspect and share. The AI Code of Conduct requires human oversight for critical decisions, and the security sheet describes human validation by default. These review surfaces keep users involved, but DiliTrust does not define a critical decision, automatic uncertainty thresholds or an escalation route for unreliable output.
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.
KMSC Law partner Todd Strang describes using Athennian AI to reduce time spent uploading corporate files by half, alongside smoother annual returns and bulk generation. The account names the firm, responsible lawyer, workflow and claimed improvement, but gives no measurement period, sample or timing method. Hitachi Digital separately identifies legal counsel Thao Harrington and legal project manager Susan Pimlott, with more than one hundred twenty entities centralized and implementation completed six months ahead of schedule. Its filing workflow stages suggested names and locations for legal approval. These accounts describe daily operations, while neither establishes measured performance for the newly launched Intelligence or MCP capabilities.
Consort Group's head of legal, Claire Dolley, describes DiliTrust as the central working tool for its contract activity in an account updated August 18, 2026. Legal and procurement jointly selected the system to replace fragmented contracting information. The implementation combined repository migration, approval workflows, renewal alerts, templates, signatures and required metadata fields. New sales employees receive mandatory training, and the shared system supports both sales and procurement. This account explains deployment and organizational adoption in concrete terms. It does not provide a measured before and after result or attribute a defined portion of the improvement to Lini's AI features.
Privilege and Confidentiality Posture
How client confidences are handled: attorney client privilege and work product treatment, segregation of one client matter from another, whether client data trains any model, and what the vendor commits to in writing rather than in marketing.
The AI terms treat inputs and outputs as customer data and prohibit Athennian and its model providers from training any AI or machine learning model with that material. Service terms protect confidential customer information through reasonable care, purpose restrictions and commitments from authorized personnel. Entity, folder and file permissions limit access, while administrators can exclude records from AI use. A broader license permits aggregated or deidentified service improvement, subject to the express training restriction. Privilege and work product treatment are not specifically addressed, and model provider retention is not mapped. External assistants connected through MCP also introduce the customer's separate provider arrangements.
The security sheet expressly excludes client data from training global models and confines ordinary AI processing to DiliTrust's infrastructure. The product uses granular permissions, encryption, separate data access and audit trails, while the vendor states that its team has no access to critical customer information. Customers can retain encryption keys in their own PKCS#11 compatible hardware security module. Private customer specific model training and a contractual training prohibition are not set out in a customer agreement available before procurement. DiliTrust does not address legal privilege preservation, production prompt retention or the full hosting and support processor chain.
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.
Responsibility for legal and regulatory compliance rests with the customer, and Athennian states that its service does not provide legal advice. The AI terms require appropriate human review before using or relying on outputs, and forbid presenting them as human generated. Proposed entity changes require approval in the product, while MCP cannot change the underlying records. Those controls support supervision of administrative and analytical work for legal, tax and finance teams. They say nothing about competence across every jurisdiction and do not define licensing requirements for nonlawyer users. Templates and third party content also carry no warranty of legal currency or suitability.
Lini assists legal teams with contract revisions, board minutes, summaries and matter updates. The AI Code of Conduct requires human judgment for critical decisions, and each workflow leaves users with material to inspect or edit. However, the professional boundary between that assistance and legal advice is not defined through an explicit advice disclaimer or jurisdictional duty framework. The audience includes corporate counsel, law firms and public organizations, whose responsibilities differ by role and jurisdiction. General human oversight does not specify who must supervise a generated resolution, verify a proposed clause or approve its use under the professional rules governing an engagement.
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.
Athennian's AI terms commit it to internal governance practices, including periodic evaluation and testing of output quality and reliability. They also allocate responsibility for human review, protect customer ownership of inputs and outputs, prohibit model training with customer data and require equivalent confidentiality and security obligations from model providers. Current product controls add approval before entity updates, source references, missing information responses and entity exclusions. The framework identifies actual commitments rather than relying solely on a general responsible AI statement. It does not identify a governance owner, release acceptance criteria or results showing performance differences across corporate tasks, jurisdictions or populations.
DiliTrust's AI Code of Conduct describes in-house development, synthetic and public training data, local processing infrastructure and customer model connectivity. It commits to testing and monitoring for bias, making AI processes understandable and retaining human oversight of critical decisions. Its security sheet adds human validation by default and places model processing inside the controlled infrastructure. These provisions give the governance approach substance beyond a general fairness statement. Named evaluation datasets, release acceptance criteria and results showing uneven performance across languages, legal tasks or populations are not supplied. Security leadership and machine learning roles are listed, but no complete model accountability process is published.
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.
The service agreement requires incident reporting within twenty four hours after a material security incident is determined; the data addendum sets twenty four hours after awareness for covered personal data incidents. It also provides subprocessor controls and notice subscriptions. After termination, customers have ninety days to request an export. The addendum requires deletion of personal data after that period unless law prohibits it, while broader service terms permit deletion and retain archival exceptions. Administrators can remove intake copies after filing and restrict access. No distinct schedule covers all AI conversations, provider traces and backup copies, leaving the complete retention boundary less specific than the incident commitments.
The platform combines user permissions, encryption, audit trails, watermarks, daily backups and monitoring. Each server request requires authentication and permission to proceed, and customer controlled encryption keys are available through an external hardware security module. Backups remain in the production country's jurisdiction at a separate location. Security testing includes internal reviews, weekly automated scans and an annual independent assessment. The privacy notice addresses website and business contact information, with general deletion on request. Those materials give no production prompt retention, no lifespan for backup copies, no complete service subprocessor list and no customer incident notification deadline.
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.
Customers are defended and indemnified against covered intellectual property claims involving the service, with exclusions for third party content, unauthorized combinations, modifications and misuse. Ordinary liability is capped at the preceding twelve months of relevant fees. Security, confidentiality, data processing and indemnity obligations have higher limits tied to annual fees, ranging from twice the ordinary cap to specified amounts reaching $2,000,000. Service warranties also permit termination after a breach, and deficient consulting work can trigger reperformance or a fee refund. AI accuracy remains disclaimed, including responsibility for incorrect filing and data entry. These protections do not amount to a remedy for every generated legal error.
Lini can propose agreement language, produce board minutes and summarize active matters, but a customer facing allocation of loss for incorrect output is not set out in an available subscription agreement. The website legal notice disclaims warranties and damages for website materials; it contains no AI output indemnity, application liability cap or product service level commitment. Security certifications and audit reports address controls rather than compensation after a mistaken recommendation. The terms governing enterprise use therefore remain dependent on procurement materials, with no stated public warranty or insurance commitment covering a flawed clause or inaccurate account of a board decision.
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.
A GraphQL interface supports reading and changing entities, appointments, ownership records, registrations and tasks. Setup requires an API key supplied in the request header, with separate regional endpoints and explicit access validation. The file interface uploads material to an entity or person and can create nested destination folders. iManage import and export uses the customer's credentials and selected folders or matters, through the Document Automation bundle. Read MCP instead requires the All-Access API addition, a configured role and access groups, and authorization in the connected assistant; it cannot write back. Power Automate connects governance events with enterprise systems and communication workflows.
CRM connections generate and populate contracts from Salesforce opportunities, HubSpot deals and Microsoft Dynamics data. Word and Google Docs support drafting, Teams and Outlook carry board materials and invitations, and Gmail brings communications into matter management. Registry partners synchronize entity and filing information. Signature integrations include Docusign, Adobe Sign and several European providers. DiliTrust Sign adds native simple electronic signatures with automatic storage of the signed file and completion certificate, rolling out progressively across modules. The external connections have described purposes and data flows, but administrator configuration, detailed field mappings and deployment prerequisites are not supplied as an implementation reference.
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.
Canada is the default core region on Amazon Web Services, with an American option specified in the order form without an additional residency fee. The legal portal identifies Montreal storage and Toronto backup locations for Canada, and Oregon for the American option. The newer processor list and API references also identify European and Middle Eastern infrastructure, without mapping all deployment terms or AI features to individual locations. Direct processing follows the contracted region, subject to support, investigation and third party exceptions. Google and Amazon support AI processing, and external MCP assistants operate under separate arrangements. Core storage location therefore does not determine every processing location.
The security sheet places European customers in Europe, Canadian customers in Canada, Latin American customers in Canada or Europe, and American customers in the United States. It describes regional options for the Middle East and Africa, including Morocco, and keeps backup copies within the same country as production. Ordinary AI processing uses local DiliTrust infrastructure, while customers may connect their own models through an API. Customer managed encryption keys are also available. DiliTrust gives no complete list of hosting countries, tenancy choices or processing implications for each customer selected model connection, so regional commitments do not resolve every deployment configuration.
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.
Annual independent audits back the SOC 2 Type II assurance Athennian identifies, covering security, availability and confidentiality. Its security explanation includes AES encryption, protected transport, access logging, multifactor authentication, penetration testing and incident response. The legal portal offers the SOC report and summaries of recovery, incident and penetration tests after a nondisclosure agreement. The open trust center does not expose report details, so the auditor, examination dates and precise covered services remain unresolved. References to NIST, ISO 27001 guidance and OWASP practices describe security approaches rather than additional certifications. These infrastructure controls also do not independently test the correctness of AI governance answers.
AFNOR certificates cover delivery and operation of the DiliTrust Governance Suite as SaaS under ISO/IEC 27001:2022 and ISO/IEC 27701:2019. Both run from February 23, 2026 through February 22, 2029, with the Puteaux site and remote access from Canada identified. The statement of applicability is version six dated June 28, 2024. The certificates can be obtained without a sales request. DiliTrust also claims SOC 2 Type II compliance, with that report supplied on request, and an ENS certification for Spain. Its separate security sheet describes audit practices, encryption and infrastructure; those explanations do not extend the certificates beyond their stated scope.
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 AI terms name Google Gemini and Amazon Bedrock, and the service agreement identifies Google's Vertex AI in the processing workflow. The current processor list distinguishes infrastructure, application services, support contractors and customer selected integrations. Anthropic and OpenAI appear under customer success analytics, a listing that says nothing about whether they power native Intelligence answers. Read MCP separately connects customers' external assistants. Subscribers to processor updates receive advance notice of new processors under the data addendum, but that mechanism does not promise notice of every model version change. The providers are identifiable, but Athennian gives no model versions or feature specific processing regions.
DiliTrust develops and operates its own legal models and names the suite's AI engine Lini. The AI Code of Conduct describes pre-finetuning and fine-tuning commercially permissible open models with synthetic legal material, public datasets and an internal knowledge base. Processing takes place within local infrastructure, and customers may attach their own language models through an API. The underlying model families and versions are not identified, despite broader marketing language about fully proprietary technology. Nor is a model specific advance change notice promised. The architecture is described, while the precise foundation model and any customer selected provider remain configuration questions.
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.
Essentials starts at $25,000 annually with one hundred entities. Professional includes two hundred entities and Enterprise three hundred, with prices supplied through sales. Additional capacity comes in bundles of fifty entities. Intelligence, AI Core, enterprise authentication and other functions can add charges; the relevant API and document connections have separate entitlements. AI action credits expire annually, and the service terms set overage at $2 per additional unit, with skill consumption subject to change on thirty days of notice. Implementation covers configuration, migration, integrations and training but has no standard fee. The entry rate is therefore a platform starting point rather than a complete AI deployment price.
Pricing depends on user numbers, the selected solutions and customization requirements. DiliTrust offers tailored arrangements and demonstrations, but no subscription rate or minimum purchase is stated. Onboarding covers setup, migration, training and support; the main offering includes ongoing multilingual support and user training. Implementation duration varies with project complexity, and consulting is available for integrations and configuration. Its ROI calculators estimate potential savings without establishing the actual invoice. A price comparison therefore requires a proposal defining modules, user scope, services and signature requirements; DiliTrust publishes no renewal terms, implementation charges or cost for optional model connections.
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.
Athennian supports corporate legal, tax, treasury, finance and compliance teams, alongside law firms and other professional advisers. Its private markets workflows cover funds, special purpose vehicles, portfolio companies and layered ownership, with clients able to review selected records and approvals. Governance work includes appointments, authority, equity and debt, minute books, reporting and regulatory dates. The company identifies more than five hundred thousand entities across over one hundred fifty countries. The figure measures operational reach, not uniform filing availability or legal research coverage in every country. Detailed boundaries for supported legal advice, practice specific accuracy and jurisdictional automation remain outside the stated coverage.
Corporate legal departments, general counsel, boards and company secretaries are the main audience for the governance suite. It combines entity administration, contracts, disputes and matter tracking, meeting workflows and protected file exchange. Industry material addresses pharmaceuticals, private equity, real estate and energy. The FAQ additionally identifies law firms and public sector organizations and says smaller businesses can use tailored packages, so coverage is broader than large corporate departments alone. Multilingual operation and international deployment are supported, but a precise jurisdictional legal content inventory, government deployment requirements and exclusions by practice type or organizational size are not supplied.
The 12 legal signals, side by side
Recorded rather than graded. These are the questions a practitioner has to answer before a tool touches a client matter, and the answers are taken from public material only.
Client Data in Training
Can material a lawyer puts into this product be used to train a model?
The April 2026 AI terms explicitly prohibit Athennian and its model providers from using customer data to train AI or machine learning models. Inputs and outputs are included in customer data. The general service agreement separately permits aggregated or deidentified analysis to improve services and other offerings, but the specific AI restriction supplies an express training boundary. Customer chosen external assistants reached through Read MCP remain a separate arrangement; the Athennian prohibition does not itself rewrite those assistants' customer contracts or retention settings.
Client material is excluded from training global models by the security sheet. The AI Code of Conduct separately describes synthetic data, public datasets and an internal knowledge base for the vendor's own model development. This is an express policy assurance for shared models, without a matching customer contract available before procurement or a clear position on private customer specific adaptation. The privacy notice's general service improvement language concerns business and user information and does not itself specify model training.
Prompt and Output Retention
How long does the product keep what a lawyer typed, and can that be set to zero?
AI inputs and outputs fall within customer data, but no distinct conversation or provider retention period is specified. The service agreement preserves a ninety day export request window after termination and permits later deletion, while the data addendum requires deletion of covered personal data after that period unless law prohibits it. Archival exceptions remain relevant. Administrators can delete Centralize intake copies after filing, but that setting addresses uploaded files rather than establishing zero retention for prompts, answers or external assistant histories.
The AI processing path ends with results returned to the user, without a stated period for keeping prompts, generated text or meeting recordings. Daily infrastructure backups are described, without a deletion cycle. The website privacy notice allows retention for as long as needed and deletion within a reasonable time on request; it sets no rules for production AI history. No customer setting for prompt expiry or zero retention is identified for Lini's contract, board or matter workflows.
Ethical Walls and Matter Segregation
Does retrieval respect the firm’s ethical walls, or can the model read across them?
Access runs on Athennian's own roles, access groups and entity permissions, with additional restrictions for individual folders and files on eligible plans. Intelligence respects those controls and permits entity or file exclusions from AI access. Read MCP applies the configured Athennian role and groups to live retrieval, so an external assistant cannot reach material its user lacks permission to view. Centralize administrators retain visibility over uploaded intake material.
These controls are specific, though Athennian describes no automatic inheritance of a law firm's external document system walls.
Administrators define roles, permissions and access to individual files and functions within DiliTrust. Server requests are authenticated and authorized before execution, with audit trails capturing user activity and changes. Watermarks and controlled sharing help track distribution of sensitive corporate material. This is the suite's own access system rather than a stated inheritance of external document repository permissions.
The materials do not specify automated conflicts screening, matter wall synchronization or exactly how every Lini retrieval request applies an individual user's file permissions.
Third Party Request and Subpoena Notice
If someone subpoenas the vendor for a firm’s data, does the firm hear about it first?
The service terms require prompt notice of legally compelled disclosure and other third party requests for customer data unless law prohibits notice. Athennian says it will reject requests it is not legally required to satisfy and attempt to redirect valid requests to the customer. Confidentiality provisions also provide advance notice where lawful and assistance with a challenge at the customer's expense. The data addendum repeats notice protections for personal data.
No public transparency report supplies request totals, challenge outcomes or a record of how those commitments have operated.
The privacy notice permits cooperation with legal process and other requests considered necessary to protect rights or prevent harmful activity. It also says the company rejects demands it considers unfounded, overbroad or unrelated to their purpose. There is no corresponding commitment to notify the affected customer, offer time to object or report the demand afterward. Regional hosting and claims about foreign government access address where information is held, not whether customers are told when a competent authority requests disclosure.
Primary Law Corpus Provenance
Where does the law in this product come from, and does the vendor have the right to use it?
Intelligence draws on corporate minute books and other customer maintained records, with direct links to the sections supporting its answers. Read MCP retrieves current entity, ownership, appointment and filing information from the platform. Those sources support governance analysis but are not an identified primary law research collection. Global compliance and beneficial ownership guides are separate resources, and the number of countries represented by customer entities does not identify an underlying legal corpus. No comprehensive inventory, rights basis and update schedule for legal authorities is specified.
Lini primarily analyzes customer agreements, board materials and matter information. Its model development description names categories of training material, including synthetic data and public datasets, but identifies no specific collection of primary law, publisher or statutory source. The statement that base models permit commercial use is not a license for a defined legal research corpus. International and multilingual capabilities likewise say nothing about jurisdiction by jurisdiction authority coverage or the update schedule behind an answer containing a legal proposition.
Good Law Verification
Does the product tell you when the authority it just cited has been overruled?
Ownership, authority, registrations and the corporate material supporting those records are all open to analysis. Intelligence links answers to source files, and MCP can identify expired registrations or incomplete records. No integrated judicial citator or subsequent treatment check is described. A registration date comparison or cited board resolution concerns the customer's governance state rather than whether a judicial authority has been overruled.
The legal guides and compliance workflows therefore include no automatic check on the continuing validity of case law used in legal advice.
Risk Detector compares contract provisions with internal rules, while other AI functions produce meeting minutes and matter summaries. No licensed citator or mechanism for showing whether a case has been reversed, distinguished or overruled is specified. Entity registry integrations keep corporate filings aligned with official sources, which serves a different purpose from validating case authority. The suite's review controls support inspection of customer materials, but they include no current-law check for legal propositions in a generated response.
Refusal and Uncertainty Behavior
What does the product do when the answer is not in the corpus?
Athennian Intelligence says it identifies missing information rather than guessing when the answer is absent from the minute books. Its cited answers let users inspect the material that does support a response, and entity selection can narrow the available context. This is an explicit no answer behavior, although no public evaluation establishes how reliably it triggers. The statement applies to native Intelligence; an external assistant connected through Read MCP has its own response behavior and cannot be assumed to follow an identical abstention rule.
Users can adjust contract suggestions, edit generated minutes and inspect a QuickView summary before sharing it. The Code of Conduct requires human oversight for critical decisions, but no AI refusal condition, confidence cutoff or request for clarification is specified for unsupported answers. A reviewer may see the original clause or meeting material without receiving a warning that the model lacks adequate support. The prescribed human role therefore leaves open how Lini itself behaves when a source is incomplete or ambiguous.
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 contained no entry naming Athennian or Paper Interactive as the responsible tool as of October 16, 2026. Neither name appeared incidentally in the reviewed database material. That finding neither shows reliable output nor measures error frequency, particularly for internal corporate administration that may never reach a filed court record. Native Intelligence and external assistants connected through MCP also have different response paths. Users still need to inspect the cited records and confirm conclusions before using generated analysis.
No fabricated citation proceeding naming DiliTrust or its Lini engine was identified in the AI Hallucination Cases database as of October 16, 2026. Matches containing the letters in Lini referred to unrelated names and ordinary words, rather than this product. Most suite outputs concern agreements, board minutes and internal matter updates, where a mistaken detail may never produce a court citation incident. This result supplies no measured accuracy assurance and does not exclude other disputes or operational mistakes involving generated content.
Bar Guidance Alignment
Has the vendor engaged in public with the ethics opinions its buyers are bound by?
Athennian's terms require human judgment before relying on AI output and prohibit describing it as human generated. The service agreement also excludes legal advice and leaves regulatory compliance with the customer. These product and contract boundaries address use of the system, but no named bar opinion or professional responsibility framework is connected to its features. Corporate ownership guides concern a different subject.
Client consent, supervision, competence and confidentiality duties therefore remain questions for the firm's own professional framework rather than an Athennian ethics mapping.
The governance framework discusses GDPR, the European AI Act, bias mitigation and human oversight. It does not name a bar opinion or connect particular controls to jurisdictional duties governing lawyers' competence, confidentiality, candor or client communication. Corporate governance obligations and professional conduct rules apply to different actors. Lawyers using proposed contract clauses or generated matter summaries therefore receive general supervision principles, without a vendor supplied mapping to the rules that govern their own engagement or professional status.
Billing and Fee Posture
Does the vendor address what happens to the bill when the work takes an hour instead of six?
KMSC Law reports that AI reduced time spent uploading corporate files by half and made annual return work faster. Athennian also records platform changes and offers AI cost and usage visibility. Neither feature establishes a client invoice entry identifying AI assistance, actual lawyer time or the treatment of saved effort. The subscription's own usage charges are separate from the legal fee charged to a client. No fee guidance explains how a firm should disclose or bill the faster entity work described in that customer account.
DiliTrust promotes reduced manual work and supplies ROI calculators for its suite and individual modules. Its stated customer audience includes law firms as well as corporate departments, so time savings can arise within a lawyer and client fee relationship. No mechanism allocates those savings to billable entries, adjusts invoices or explains AI charges to clients. Matter management tracks outside counsel spend and dispute costs, which is a different function from an AI billing disclosure process or a client facing account of automated work.
Outside Counsel Guideline Readiness
Can a firm get this vendor through a client’s AI clause without a bespoke negotiation?
Athennian supplies a current processor list with names, purposes and locations, together with AI terms identifying native model providers and a contractual training prohibition. The legal portal also provides service terms, a data addendum and security information. Detailed audit reports require a nondisclosure agreement. No dedicated client consent or notification package is identified, and external MCP assistants add the customer's chosen providers to the assessment.
The available provider information supports initial diligence without itself implementing each client's outside counsel restrictions.
An AI Code of Conduct, a security sheet and downloadable ISO certificates provide materials that legal teams can share during a client technology review. They describe local processing, model development, global model training restrictions and security controls. Customers may also connect their own models, which changes the configuration requiring review. The website privacy notice lists business service providers, but no complete production subprocessor schedule covers hosting, monitoring and support. A client specific approval workflow or automated mapping to outside counsel guidelines is not included.
Court Disclosure Support
If a judge’s standing order requires an AI disclosure, can the product produce one?
Enterprise administrators can view and export the Global Audit Trail as an Excel file, including changes to entities and people, access events and API activity. Entries identify dates, users, affected records, locations and change details. AI source citations and mandatory approval of proposed updates provide additional review context. No combined output is described that joins every generated passage with its exact model version, complete retrieval context and a named human verification step. Regulatory or court disclosure may therefore require reconstruction beyond the exported application activity.
Risk Detector leaves highlighted source clauses and tracked suggestions for human review, while the suite maintains activity and change logs. These supply parts of a reconstruction of contract work, and signed files have their own completion certificate. No per-file export combines the model version, prompt, retrieved sources, generated passages and reviewer approval into an AI disclosure history. Board minutes and QuickView updates may be shared after editing, but no tribunal or regulator declaration identifies which portions originated with Lini.
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.
- Primary Law Corpus Provenance
- Good Law Verification
- Bar Guidance Alignment
Product updates
The latest logged changes from each vendor, newest first, each linked to its source.
Athennian
-
Product / capability
Admins can now build dedicated entity pages for teams such as Tax, Banking and Finance, combining custom records and fields with granular role based access.
Source -
Product / capability
Athennian launched Athennian Intelligence, which answers plain language questions over corporate records and minute books with links to the source and returns structured results to Custom Reports.
Source -
Product / capability
Athennian's 25 Sept release lets organizations decide whether pending transactions count in ownership calculations and structure charts, with ownership views now showing issued, pending, reserved and available amounts.
Source
DiliTrust
-
Privilege / confidentiality
DiliTrust published documentation that separates where customer data is stored from where AI requests are processed, and names four processing options.
Source
In summary
Athennian
Athennian, from Paper Interactive, Inc. of Calgary, Alberta, doing business as Athennian, is an entity management platform for in house legal, tax, treasury and finance teams at corporate groups and private markets firms, and for the professional firms that serve them. It holds entity and people records, ownership structures and charts, appointments, documents and compliance tasks across jurisdictions. Its AI extracts governance data from uploaded documents and answers governance questions under human review. According to the AI Legal Index, its published AI terms bar training on customer data and name Google and Amazon as model providers. Its case studies and pricing sit on separate pages.
DiliTrust
DiliTrust, from DiliTrust SAS of Paris La Defense, sells in house legal departments and corporate boards an integrated suite of five modules: a board portal, contract management, entity management, matter management and a dataroom. Its AI engine, Lini, drafts board minutes, transcribes meetings, extracts contract terms, summarizes documents and flags risky clauses. According to the AI Legal Index, DiliTrust publishes its ISO certificates as downloads alongside an AI code of conduct, states that its models are its own and run in regional data centers, and names customers including BNP Paribas and LVMH. Organizations can connect their own language models through its API.
Questions buyers ask
Athennian vs DiliTrust: which is better for entity management?
They are close, and the difference is scope. Athennian is a dedicated entity management platform with published AI terms and a two way API. DiliTrust adds a board portal, contracts and matters in one suite and publishes its certificates. Teams that want terms in writing have more to read from Athennian. From the AI Legal Index, based on each vendor's own published materials as of September 26, 2026. No vendor pays for placement.
Does Athennian train AI on customer data?
No, by contract. Section 9 of Athennian's AI terms, updated 14 April 2026 and part of its services agreement, states that Athennian does not use, and does not permit its model providers to use, customer data to train any AI or machine learning model, and it defines input and output as customer data. DiliTrust describes its training data as synthetic, public and its own, with no customer agreement published. From the AI Legal Index, based on each vendor's own published materials as of September 26, 2026. No vendor pays for placement.
Is DiliTrust subject to the US CLOUD Act?
DiliTrust's AI code of conduct states that its AI solutions are not subject to the US CLOUD Act, except for US clients, and that its data centers sit in the regions where clients operate. Its privacy policy says it will refuse government requests it considers unfounded or too broad. Nothing published commits it to tell a customer when its data is demanded. From the AI Legal Index, based on each vendor's own published materials as of September 26, 2026. No vendor pays for placement.
Which AI models do Athennian and DiliTrust use?
Athennian's AI terms name Google Gemini and Amazon Bedrock among its model providers and point to a published subprocessor list for others. DiliTrust states that its models are designed and trained by its own teams, fine tuned from open source and commercially permissible models, and lets organizations connect their own language models through its API. From the AI Legal Index, based on each vendor's own published materials as of September 26, 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.
Three points to weigh. DiliTrust publishes no customer agreement, so its commitments on training, data handling and liability rest on a code of conduct and product pages rather than a term a buyer signs. Its models are fine tuned from open source models it does not name. Athennian's liability cap sits in its service terms, and its price sits on a separate pricing page. Neither vendor reviewed this page.