A

Aline

The models drive the capabilities the vendor leads with, on a contract system that works without them, which is the B band. Aline's AI drafts agreements and memos from a team's clause library, redlines third-party paper against playbooks generated from the team's templates, answers questions about one contract or thousands, and extracts terms from the repository without tagging. The vendor itself says the contract infrastructure came first and the AI layer was added to a system that already understood the documents: approvals and intake workflows, native e-signature through AlineSign, a document repository and file-storage sync all function without the models. That is the B description exactly. Verified 18 September 2026.

Accuracy and grounding are asserted without measurement or a described method, which is the C band. The Legal AI page says every output cites the clause, source or precedent it came from, that research answers are grounded in the customer's contracts, statutes and approved legal sources and never a hallucinated case, and that the product reads full-length contracts rather than a slice. None of this is measured: no accuracy figure, test set or evaluation is published, no failure mode is named, the approved legal sources are not identified, and nothing describes what the system does when it cannot support an answer.

The published terms run the other way, stating in section 9(d) that outputs are not tested, verified, endorsed or guaranteed to be accurate and that customers should independently verify all outputs. The distance between the marketing and the agreement is the finding. Verified 18 September 2026.

A written position that judgment stays with the lawyer, with real review surfaces, short of a control structure, which is the B band. The about page states that Aline does not make final calls or replace in-house counsel's read on a deal, and that legal sets the playbook and everyone else works inside it. The product routes contracts for approval on triggers such as value, department or clause-level risk and keeps an audit trail of decisions; redlines are proposed against playbooks for a lawyer to accept.

The terms require customers to verify all outputs and bar using the AI for automated decisions with legal or similarly significant effects without adequate human review. What is missing for A is a threshold or categorical limit on a named mode: sales users can launch agreements without legal inside the guardrails, and nothing states which outputs may go out without a lawyer's review. Verified 18 September 2026.

Named customers with figures, short of any method, which is the B band. The site attributes results to named people at named companies: Xevant's senior legal director says a mutual NDA redline fell from up to 45 minutes to under two, a PharmaCann attorney says a 50-page commercial agreement's first pass takes 15 minutes, and testimonials name Muon Space, Telly, 7th Level, Bison Transport and DailyPay, with a linked 7th Level case study.

The about page adds aggregate figures, 95 per cent faster review, $40,000 a year in outside counsel spend saved and 8,400 contracts queried by one team, and says 2,000 users across more than 100 companies. None of the figures states how it was measured or over what period, so they are testimonial figures rather than outcomes a reader can assess. Verified 18 September 2026.

A contractual training bar and a named model supply, short of confidentiality terms, segregation and privilege, which is the B band. Section 9 of the Terms of Service states that Aline shall not use inputs or outputs to train its AI features, names OpenAI and Anthropic as the providers behind them, and leaves intellectual property in inputs with the customer; the security page adds that agreements with the model providers legally prevent training on customer data.

The agreement has no general confidentiality clause protecting customer contracts, the privacy policy permits use of aggregated or de-identified information for any purpose and archiving after removal, privilege and work product are not addressed, and the only separation statement is that queries run inside the customer's secure contract environment. Section 1 states that no attorney-client relationship or privilege is created with Aline, which concerns Aline's own role rather than the customer's privilege. Verified 18 September 2026.

A clear disclaimer sits in the terms while the marketing addresses non-lawyers in advice terms, which is the C band. Section 1 of the terms is plain: Aline is not a law firm or an attorney, nothing on the service is legal advice, no attorney-client relationship is created, and anyone needing legal advice should consult a licensed attorney. Against that, the product is sold to sales, procurement, finance and founders as well as lawyers, the home page promises accurate documents drafted without expensive counsel, and the FAQ says business users can launch agreements without deep legal expertise.

The vendor's line that legal sets the playbook and others work inside it is a real guardrail, but nothing addresses supervision or competence duties for the lawyers who set it, and no jurisdiction limit is stated. Verified 18 September 2026.

AI use rules are incorporated into the agreement, short of any evidence of how the vendor builds and tests, which is the B band and the ceiling for customer-side governance. Section 9(f) of the terms prohibits using the AI for automated decisions with legal or similarly significant effects on individuals without adequate human review, for discriminatory or harmful purposes, or to represent output as vetted by Aline or as wholly human-generated, and the about page publishes principles including that judgment stays human and that claims should be earned.

What is missing for A is the vendor side: no one inside Aline is named as accountable for model behaviour, nothing describes testing before release, and nothing is disclosed about uneven output. Verified 18 September 2026.

Security basics are published while the privacy policy covering customer data is generic, which places this at C. The security page states encryption in transit and at rest, regular penetration testing, single sign-on through Microsoft Entra and Okta, and hosting on AWS in the United States, and the pricing FAQ says customers own their data, can download it at any time, and have a grace period after cancelling before it is removed.

The privacy policy, last revised in November 2023, retains information as long as needed, permits keeping data in archives after removal from the service, names AWS, Sendgrid and Stripe among providers without a full list, and states no incident notification commitment; nothing addresses how long prompts and outputs are kept. Verified 18 September 2026.

A published liability position with a vendor indemnity and a cap, short of anything that stands behind output, which is the B band. Section 14 of the terms gives a vendor indemnity against third-party claims that the service infringes intellectual property, with modify, license or refund remedies and stated exclusions; section 15 caps Aline's liability at fees paid in the prior twelve months and excludes indirect damages for both parties.

Section 9 then excludes infringement claims arising from outputs, states that outputs are not tested or guaranteed accurate, and applies the service disclaimers to the AI features. No insurance position is published. Verified 18 September 2026.

The principal integration is documented to implementer depth in the vendor's public help centre, which is the A band. The help centre's Salesforce managed package guide sets out who installs what, the install scopes, the Aline Admin permission set, OAuth pairing in which Aline writes a signing key into the Salesforce org and signs every request with it, revocation on disconnect, which Opportunity, Account and Owner fields flow into documents, how editable and read-only fields behave at launch, and an error-log tab with named failure cases.

A companion article documents mapping CRM fields from Salesforce or HubSpot into template fields, with override at launch. The pricing and product pages name file-storage sync with Google Drive, Dropbox, Box, OneDrive and SharePoint, CRM integrations with Salesforce, HubSpot, Pipedrive and Close, approval notifications in Slack, and single sign-on through Microsoft Entra and Okta; the storage sync is not documented to the same depth.

No document management system used by law departments, such as iManage or NetDocuments, is named. Flagged for sampling. Verified 18 September 2026.

The hosting model and region are stated, with no options and no tenancy detail, which is the B band. The security FAQ states that Aline runs on AWS in the United States, and the terms and privacy policy confirm storage and processing in the United States, while also reserving processing in any country where Aline maintains facilities. No single-tenant or private deployment is offered, no other region is available, and processing location for calls to the model providers is not stated separately. Verified 18 September 2026.

A named attestation is claimed, short of any route to the evidence, which is the B band. The security page states that Aline is SOC 2 Type II certified and regularly penetration tested. No auditor, report period or scope is given, no trust centre exists, and no route to obtain the report is described beyond contacting the team. The same page describes Aline as GDPR certified, which is not a certification scheme, and the pricing FAQ uses the word compliant for the SOC 2 claim; both are recorded as imprecision in the claims rather than as findings against the attestation. Verified 18 September 2026.

Model providers are named, including in the agreement, without any change notification, which is the B band. Section 9(e) of the Terms of Service names OpenAI and Anthropic as the providers behind the AI features, and the home and about pages say the product runs on Anthropic Claude, OpenAI and Google Gemini, picked per task. The contractual list, last revised in November 2023, no longer matches the site, which adds Google.

No model version is given, and the terms reserve the right to modify the service without notice, so no commitment exists to tell customers when a provider changes. Verified 18 September 2026.

A buyer can price the product before a sales conversation, including what implementation adds, which is the A band. The pricing page publishes a Team plan starting at $200 per full user per month and $50 per support user per month, with unlimited documents, workflows, templates and AI reports, playbooks and chat, and up to 20 hours of onboarding, implementation and training included; the FAQ states there are no maintenance or contract-volume fees and offers a 21-day free trial.

Enterprise pricing is by quotation, and the page's annual figure is shown as $24,00, a display error. A pricing row is written. Verified 18 September 2026.

The segment is defined precisely and the contract work described with substance, short of practice boundaries, which is the B band. The vendor says it is built for two-to-five-person in-house legal departments, and for GC-led teams at companies of 100 to 1,000 employees, contrasting itself with enterprise CLMs built for 50-person departments; it also serves procurement, finance and sales users and has recently added a law-firm offering.

The work covered is commercial contracting: NDAs, MSAs, DPAs, vendor agreements, policies and board memos, plus legal research questions. Nothing states which jurisdictions or contract types it handles less well, and the scope of the research function is not bounded. Verified 18 September 2026.