Josef vs Legito: how they compare in 2026

Josef sells three products: Josef Q for AI question and answer, contract automation, and workflow automation. Only the first is model-driven. The contract and workflow modules are template and rules automation that predate the AI layer, and they are what the L’Oréal case study’s 66% turnaround figure actually rests on. Josef Q is not peripheral — it has its own login at q.joseflegal.com, four dedicated product pages and its own named customers at adidas, Bupa and Cityblock Health — but remove it and a working no-code document and workflow platform remains. That is the B band precisely: the models are the engine of a core capability layered on a product that would still function without them. Pages read 1 September 2026.

The vendor's own headline settles this: the AI page is titled AI-powered Document Automation and CLM and its subtitle reads that AI-powered and rule-based document automation complement each other. Kedy AI is a real, named, shipped layer with five distinct capabilities, but it sits on top of a platform that stands entirely without it. Remove the models and a buyer still has interactive template automation, end-to-end lifecycle management with routing and approvals, a no-code application builder, an audit trail and a free electronic signature product, which is the whole of the free plan and the whole of what the named customers are quoted describing. The company was founded in 2015 and its case studies, including the PwC deployment and the 450,000 hours figure, rest on rule-based automation rather than on inference. AI accelerates authoring and extraction here; it is not the mechanism the buyer is paying for. Checked 7 September 2026.

Grounding is real and specifically documented. The AI controls page states Josef Q is a closed-domain system answering only from uploaded content, and the source list shows the exact page, paragraph and clause an answer drew on, so a reader can open and check the support. A document pre-processing engine is described as scanning and optimising uploaded content before indexing. No measured accuracy figure and no test set is published anywhere, which is what holds this at B. Two limbs of the higher bands do not apply to this product class: Josef Q retrieves the customer’s own policies rather than primary law, so grounding to primary authority and citation-status checking are not capabilities it claims. Against that, the audit trail page carries an unqualified “No hallucinations!” claim, which is marketing running ahead of the mechanism. It does not drop the grade to D because it sits alongside a described architectural control and a published refusal path rather than standing alone. Read 1 September 2026.

No accuracy content of any kind is published. There is no figure, no test set, no evaluation, no third-party validation and no named failure mode across the AI page, the security page, the pricing page or the subscription agreement, and the agreement does not mention AI at all. The nearest thing to a control is a product-design constraint: AI-powered drafting is described as inserting extracted data into automated templates while maintaining 100 per cent integrity of the template, which bounds what the model can alter but is a statement about template mechanics rather than about output accuracy. The nearest thing to a limitation disclosure is a blog article that asks whether a reader can accept that AI returns different results in two identical situations, which acknowledges nondeterminism in the abstract without stating what the product does about it. Grounding to primary legal authority does not bite on a document automation and CLM product and is counted neither way, per the inapplicable-limb rule. The knowledge base article on AI document drafting was not opened and is treated as corroboration only; the grade rests on the absence of accuracy material across the surfaces that were read. Checked 7 September 2026.

The control structure is published in unusual detail for this lane. The AI controls page documents human-in-the-loop moderation over every answer, source addition and removal in a Moderation tab, a Tuning Lab for tone, length and spelling convention, and Smart and Follow-up Suggestions that shape what end users ask. The stop rule is published and specific: outside the uploaded corpus the tool answers “Sorry, I don’t know.” The audit trail page adds that all user questions and generated answers are tracked and stored. What is missing for A is the route back to human judgement — no escalation path from an unanswered question to a lawyer was located on any product page read on 1 September 2026 — and no thresholds or distinct autonomy modes are published. B rather than A on the absent escalation route, not on the review surfaces, which are the strongest part of this record.

Autonomy is claimed and oversight is not described. The vendor states that AI agents will undertake regularly recurring tasks in the customer's workspace for them, which is unattended operation, and that Kedy AI navigates through workflows and controls lifecycle operations. Nothing published states what an agent may do without a human, at what point it stops, whether a confidence signal is shown to a reviewer, or what happens after an agent acts wrongly. No approval gate specific to AI output is described anywhere. The platform-level audit trail, described on the security page as timeline and versioning, is a general control that predates the AI and is not presented as an AI oversight surface, so it is named here rather than credited as one. The subscription agreement, which is where a supervision obligation would ordinarily sit, is silent on AI entirely. Checked 7 September 2026.

Attribution and figures are both present, which is more than most of this pull manages. The L’Oréal case study names Candy Welsh, Legal Counsel, and reports a 66% drop in contract turnaround, one contract moving from over an hour to 20 minutes, 100-plus such contracts a year, and roughly two weeks of work saved annually. Further named deployments carry named officers: Katherine Roseveare, General Counsel at adidas, across 65,000-plus people; Claire Nuske, Head of Legal Operations at Bupa, with 40-plus self-service tools; Wendy Chow, General Counsel at Cityblock Health; Sheila Dusseau, Head of Global Legal Operations at Ferring Pharmaceuticals. What keeps this at B is that no case study read on 1 September 2026 carries a date, and the 66% figure is given with no statement of how it was measured or over what period, so a reader can assess neither the method nor the currency.

Deployment evidence is unusually broad and well attributed. Named customers appear with logos including PwC, Deloitte Legal, LexisNexis, Holland & Knight, Santander, Société Générale, UniCredit, Telia, Škoda, Vialto, U.S. VETS and the Bronx and New York County District Attorneys' offices, against a stated base of over 400,000 users. Testimonials carry full names and roles rather than initials: Tomas Fiala, Director at PwC, describing a Centre of Excellence running since 2018 across the PwC global network; Paul V. Saunders, Chief Innovation Officer and Partner at Stewart McKelvey; Sony Varughese, Chief Technology Officer; Nel Pelle, Sales Manager at BAM; Galya Oster at Atrafin; Eli Cohen, Chief Operating Officer at RBI; and Jeremy Aber, a software attorney. Figures are specific, including contract checking falling from 20 to 50 minutes down to 5 minutes, 450,000 hours saved at a Big Four firm, and a 10,000-user deployment built by two part-time specialists. Two things hold this at B. The figures sit inside testimonials and case-study titles with no method, baseline, sample or date. And none of the outcome evidence is attributed to Kedy AI: every quoted result describes rule-based document automation, so a buyer evaluating the AI specifically has adoption evidence for the platform and none for the models. Checked 7 September 2026.

No customer agreement is published. The home page, security page, FAQ, website terms and privacy policy were read on 1 September 2026 and a clause-language search was run; the only two published instruments are website terms and conditions and a privacy policy dated 15 February 2024, and that policy states it applies to Josef as a controller rather than to customer content held as a processor. What is published is real but general: the FAQ states customers retain full ownership of their data and Josef acts as custodian, and the security page states AES-256 at rest and TLS 1.3 in transit with Josef-held server-side keys. Nothing addresses training on customer content, tenant separation, or privilege and work product treatment. The in-house band amendment does not rescue this, because for an in-house buyer the requirement is tenant-level separation and no separation of any kind is documented. C rather than D because the ownership and encryption commitments are published and readable before signing; C rather than B because the commitments that decide this axis cannot be read in advance at all.

Substantive published commitments across several limbs, short of the full picture. The subscription agreement makes all data, files, documents and templates uploaded or created by the customer the customer's property, and grants Legito the right to use customer data solely for the purpose of performing under the agreement, which is a purpose limitation rather than a broad licence. Section 5 makes customer data the customer's confidential information and binds Legito to reasonable care, purpose limitation and need-to-know access under signed confidentiality agreements. The AI page adds that no third-party service processes customer data in AI and that data is never used for AI training. Retention is bounded: the customer may export for 30 days after termination, after which Legito may destroy the data. Segregation is real, with the customer choosing a named storage region and the processing agreement barring transfer outside it without express written consent. Encryption is applied in transit and at rest with keys held away from the hosting provider, and the product ships a self-anonymisation feature for data in documents. What is absent is the limb this axis exists for: privilege and work product are not mentioned anywhere in the agreement or on any published surface. The product holds contract and back-office documents rather than matter files, which makes that omission less costly to most buyers than it would be for a litigation product, but the limb is required and it is not met. Checked 7 September 2026.

The only disclaimer located is scoped to the wrong thing. The website terms and conditions state that content on the Josef website is provided for information purposes only, which addresses the marketing site rather than the output of a tool answering policy questions for employees. Meanwhile the product is described in advice terms: the Josef Q demo panel offers to provide scalable guidance and advice, and the FAQ describes legal aid organisations and boutique firms using Josef to deliver client-facing onboarding, wills and guidance tools to members of the public. No jurisdiction limit, no statement of who may rely on an answer, and no consumer-facing disclosure requirement were located on any page read on 1 September 2026, and ABA Formal Opinion 512 is not mentioned. C rather than D because a disclaimer does exist in the terms; C rather than B because it does not reach the product, and the client-facing uses the vendor advertises are exactly where the advice line bites hardest.

Nothing addressing unauthorised practice, professional responsibility or the boundary between the software and legal advice was located on any Workspace surface. The subscription agreement, which is the customer agreement for the product indexed here, contains no such provision, and neither the AI page nor the legal department page nor the law firms page carries a disclaimer. One instrument on the same domain does address it: the Legito Marketplace terms of service require a user to tick a box agreeing that the application is not a legal service or its substitute. The Marketplace is a separate product with its own purchase path, its own terms and its own consumer-facing single-use conditions, so that acknowledgement governs a different transaction and is not credited to the Workspace record. It is named here because a reader who finds it unaided should see that it was considered and why it was set aside. The grade records what is establishable on the date and the cause sits in this note rather than in the letter. Checked 7 September 2026.

No governance position located. The home page, the three product pages, the AI controls page, the audit trail page, the security page, the FAQ, the website terms and the privacy policy were all read on 1 September 2026. There is no responsible AI page, no named internal owner, no pre-release testing regime, and no ISO 42001 or equivalent. Nothing addresses whether answer quality varies across topics, document types or populations. The moderation and review controls are real but they are an oversight mechanism already graded on Autonomy, and the band excludes security controls as a different subject, so neither can be spent here. The chief executive’s published article on context engineering is a design argument, not a governance disclosure.

No governance disclosure was located. There is no AI policy, no acceptable use policy for AI, no responsible AI or ethics statement, no bias or fairness discussion, no ISO 42001 or NIST AI RMF alignment claim, no named internal owner for AI decisions, no model evaluation or testing description and no AI-specific section in the subscription agreement or the data processing agreement. Two blog articles touch on AI in document-centric processes and on where AI brings value and where its limits lie, but both are market commentary published to the blog rather than statements of how the vendor governs its own models. The ISO 27001 certification covers information security and its quoted scope names document automation and CLM software development and hosting, not AI governance, so it is not credited here. Surfaces read on 7 September 2026: the AI page, the security page, the subscription agreement and its data processing exhibit, the pricing page and the site footer inventory.

Access control is documented and the rest is not. The security page states AES-256 encryption at rest, TLS 1.3 in transit, Josef-held server-side encryption keys and regular access audits, alongside OWASP and ASD secure coding frameworks, automated vulnerability scanning and annual third-party penetration testing. What happens to uploaded documents and prompts afterwards is not addressed: the privacy policy retains service data for the duration of the business relationship and an unstated period after, deletion is described only as securely deleting or anonymising once no legitimate business need remains, and no incident practice is published anywhere. The subprocessor list is not published — the privacy policy says it can be provided on request, and no self-service portal or instant-download route was located on 1 September 2026, so under the three-tier rule it earns no credit. C rather than B because two of the five elements the band names are absent outright and two more are stated without a period or a mechanism.

Three express commitments sit together on the AI page and they are the strongest AI-specific disclosure this vendor makes: Kedy AI is hosted 100 per cent on Legito servers, no third-party services are used to process customer data in AI, and customer data is never used for AI training. Read against the subscription agreement they are consistent rather than contradicted, because the agreement grants Legito the right to use customer data solely to perform under the agreement and reserves no training or improvement right anywhere. The one clause that comes close, the statistical information provision, permits Legito to compile and publish performance statistics about the service on condition that they neither identify customer data or the customer nor permit re-identification, which is a performance metrics right rather than a machine learning right and is not read as one. The processing agreement adds a no-transfer-outside-the-chosen-region commitment, an eight-hour breach notification undertaking and a requirement that sub-processors be bound to confidentiality. What holds this at B is placement: every AI-specific commitment lives on a marketing page, and the customer agreement, last modified in September 2023, does not mention AI at all, so a buyer has the vendor's word rather than a term to enforce. Checked 7 September 2026.

Nothing published on who bears the loss when an answer is wrong. The only liability clause located sits in the website terms and conditions and excludes liability arising from use of the Josef website, limited where it cannot be excluded to resupply of the relevant services or information. That is a website clause, not a position on product output. No master subscription agreement, customer agreement, service level agreement, warranty or indemnity was found. The escalation ladder was run on 1 September 2026: the footer was checked across six rendering pages and offers only the website terms, the privacy policy and a cookie notice, and a search on customer-agreement clause language returned no Josef instrument. This is an established absence rather than a retrieval failure, so it is graded rather than left unwritten under R7.

A published agreement with real general recourse and no AI-specific recourse at all. The machinery that exists is more than most records on this axis carry: a tiered availability warranty crediting 10, 25 or 50 per cent of the monthly fee by uptime band; a major-outage right to terminate and recover prepaid fees where availability falls below 98 per cent for two consecutive months or any three in six; warranties that security and functionality will not materially decrease and that the service will perform in accordance with its documentation; a 30-day cure period for material breach; an intellectual property indemnity covering defence costs, negotiated settlements and court-awarded damages; and an eight-hour breach notification undertaking in the processing exhibit. Against that, liability is capped at fees paid in the preceding twelve months and all indirect damages are excluded, including loss of or unauthorised access to data. The decisive point is that nothing in the agreement addresses AI output: there is no warranty, no allocation of responsibility and no remedy where Kedy AI extracts a term wrongly, drafts from the wrong source or an agent acts on a bad inference. The processing exhibit also disclaims liability for customer data disclosed through a cyber-attack notwithstanding reasonable efforts, which cuts against the recourse the rest of the agreement builds. Checked 7 September 2026.

Thirteen integrations are named and grouped by function, and the direction of travel is described: policies sync in from SharePoint and Confluence, data is pulled from Airtable and Google Sheets, and captured information is pushed out to Slack, Teams, SharePoint and Salesforce. The legal systems are what matter in this lane and they are present — HighQ, Xakia, Dazychain and Actionstep are all named. The Xakia entry goes furthest, describing Josef bots launched from Xakia’s internal client portal so that requests stay managed within Xakia. Two integrations carry linked walkthroughs, for Actionstep and Xakia. B rather than A because no documentation an implementer could use was located: no developer index, no field mapping, and no statement of what a firm must configure, with the integrations page answering any unlisted case by inviting a conversation with the team. Read 1 September 2026.

Integration is documented rather than asserted, and the documentation is public. A REST API is published with a Swagger specification, alongside webhooks described as a push API, programmable tags, a developers hub and self-service API key creation, deactivation and deletion. Named off-the-shelf integrations include Salesforce, Microsoft SharePoint and Zapier, and the subscription agreement itself names DocuSign, Salesforce and HubSpot as third-party services the platform interoperates with, which is unusual corroboration because it puts the integration surface in the contract rather than only on a product page. A Model Context Protocol server is published so that other AI applications can reach Legito features. What holds this below the top band is the absence of legal practice systems specifically: no document management system such as iManage or NetDocuments, no practice management or matter management system, no time and billing system and no court or filing system appears anywhere in the integration inventory. The depth is real and the direction is general enterprise back office rather than the legal stack. Checked 7 September 2026.

The security page and the FAQ both state data residency options in the US, Europe, the UK and Australia, a four-region commitment stated consistently in two places. The tenancy model is never stated: nothing published says whether customers sit in shared or separate instances. Processing location is not addressed as distinct from storage, and there is a tension a buyer should read before signing. While the security page offers regional residency, the privacy policy tells users their personal information may be transferred and processed outside their country including in the United States, relying on standard data protection clauses for EEA and UK transfers. The two are not necessarily inconsistent, since the privacy policy governs Josef as controller rather than the platform tenancy, but no document read on 1 September 2026 reconciles them. B on the second limb of the band: residency is offered without the processing location being addressed.

Residency is published, specific and contractually locked, which is the stronger of the two co-equal limbs on this axis. The security page states that each customer may choose their data storage location and names three: Vint Hill, Virginia in the United States; Limburg, Germany in the European Union; and Melbourne, Victoria in Australia. Hosting is with OVH, named openly. The commitment is not merely marketing: the data processing exhibit provides that Legito may not transfer customer data to a third country outside the chosen server location without the customer's express written consent, which converts the choice into an enforceable term. Two gaps sit against that. Tenancy is nowhere stated, so a buyer cannot tell whether the service is single-tenant, pooled or configurable, and no self-hosted, private cloud or on-premises option was located. There is also an inconsistency the vendor should resolve: the account creation form on the AI page offers a server choice of Australia, Canada, Germany and the United States, adding Canada to the three locations the security page lists. Both are the vendor's own current surfaces and neither is reconcilable to the other from outside, so the discrepancy is recorded and nothing is graded on Canada. Checked 7 September 2026.

SOC 2 Type II is stated on the security page, the FAQ and the home page, and the privacy policy states SOC 2 Type I and II. Annual third-party penetration testing is stated, as is GDPR compliance and ISO certification of the underlying server infrastructure. Absent is everything that would let a buyer check any of it: no auditor is named, no scope or coverage period is given, no report or summary is offered by any route, and no trust centre exists — no Vanta, SafeBase or equivalent portal was located on 1 September 2026, including through the footer and navigation of pages that render. B rather than C because the standard is named in prose rather than appearing only as a badge image; B rather than A because there is no access flow to the evidence at all, not even a gated one.

This record pulls in two directions and the note has to carry both. On the credit side, Legito does something most vendors in this corpus do not: it names its certification body. The security page states approval by Lloyd's Register to ISO/IEC 27001 and quotes the certified scope verbatim, covering the development and hosting of Legito document automation and contract lifecycle management software and the provision of document automation and software configuration services. A named auditor plus a quoted scope is A-band behaviour and is the reason this does not sit lower. On the other side, the certification is cited as ISO/IEC 27001:2013, a revision superseded by the 2022 edition, and the page states that the latest ISO 27001 audit was conducted in July 2021 on a page whose own last-modified date is September 2024. No certificate document is published, no certificate number or validity window is given, there is no SOC 2 report and no trust centre or document request portal of any kind. The OVH certifications listed on the same page, being ISO 27001, SOC 1 Type II, SOC 2 Type II, PCI DSS and a Cloud Security Alliance STAR self-assessment, belong to the hosting provider and are not credited to Legito. The penetration testing described is likewise the customers' mechanism rather than the vendor's: the page states that large customers engage third-party auditors and that Big Four firms regularly test the application, which is a disclosure about customer diligence, not about a programme Legito commissions. Checked 7 September 2026.

Josef acknowledges a large language model underneath and never says whose. The chief executive’s published position is that an LLM alone cannot be trusted and needs context engineering, and the audit trail page refers to GenAI engineering, but no model, no model provider and no inference location is named on any page read on 1 September 2026. Hosting region and ISO-certified infrastructure are published, but a hosting location answers where the software runs rather than whose model sees client content, and cannot be spent on this axis. There is no commitment to notify customers if the underlying model changes. The only route to the supply chain is the subprocessor list the privacy policy says is available on request.

Nothing about the model itself is disclosed. No model is named, no provider is identified, no model family or version is given, and the vendor does not say whether Kedy AI is built in house, fine-tuned from open weights or licensed and run on Legito infrastructure. What is published is architectural and it is genuinely useful to a buyer: Kedy AI is hosted entirely on Legito servers and no third-party service processes customer data in AI, which tells a firm that no external model provider sees client content. That fact has already been credited where it belongs, on deployment and on data stewardship, and crediting it a second time here would answer a question this axis does not ask. Where the models run is not what the models are. The three AI services Legito does name, being OpenAI, Anthropic Claude and Google Gemini, are offered expressly as optional alternatives for customers who would prefer them and the page states that Kedy AI does not require any integration, so they are a customer-elected substitute rather than Kedy's supply chain and are not counted as partial disclosure of it. There is also an unresolved tension a buyer should see: an assurance that no third party processes data in AI is difficult to reconcile with an undisclosed model of unstated provenance, and only the vendor can close that gap. Checked 7 September 2026.

No pricing information at any level. There is no pricing page: the navigation and footer were checked on 1 September 2026 across the home page, the Josef Q and integrations pages, the security page, the AI controls page and the FAQ, and none carries a pricing entry. Every commercial path is a demo booking or a contact form. No tier names, no packaging structure, no unit of charge and no figure are published, so a buyer cannot learn what is being charged for, let alone how much. D rather than C because C requires the shape to be visible with only the number withheld, and here neither the shape nor the number is published.

Real pricing information is published, but no rate is. What a buyer can establish first-party: a permanently free plan exists and its scope is named, carrying document lifecycle management, the custom application builder and Legito Sign as a lite version of the enterprise workspace, described as free of charge forever; the electronic signature product is offered at no fee; and a 30-day trial opens all features with no credit card required. Pricing pages are maintained in six languages. What is not published anywhere first-party is a paid figure, a volume band, a unit of charge, a minimum term or a seat minimum, and the pricing page carries no plan table in its rendered body. Two retrieval notes belong here rather than in the grade. The Complete Features List section of the pricing page is a heading with no content in the extracted body and is recorded as unrendered rather than absent, since the page's own structure says something belongs there. And a third-party site publishes specific per-user monthly rates and a minimum user count, which is excluded as an aggregator source and is named only so a later reader knows it was seen and refused rather than missed. The grade sits above the floor because a free tier and stated trial terms are pricing information, and below the middle band because the paid model is entirely absent. Checked 7 September 2026.

Coverage is set out by segment with named customers in each, which is unusually concrete. The FAQ and the customer navigation name five segments — in-house legal, global law firms, boutique and NewLaw firms, legal aid organisations and community legal centres, and law schools — each with its own page and named users: L’Oréal and Bumble in-house, Orrick and Herbert Smith Freehills among firms, Hive Legal and Polaris Lawyers in NewLaw, Housing Court Answers and Everyday Justice in legal aid, and Cornell, NYU and the University of Pennsylvania in education. Document types are named too — NDAs, MSAs, employment agreements, powers of attorney and client onboarding forms — with Q&A use cases grouped under commercial, privacy and data, legal operations and HR. B rather than A because the limits are not stated: the FAQ’s position is that anything repeatable can be automated, government use is not addressed, and no page says where the product stops. Read 1 September 2026.

The legal surface is real and named but sits inside a wider back-office platform. A dedicated Legal department page addresses legal operations directly, framed as no-code automation for legal ops to leverage expertise at scale with controls, automation and oversight. A separate Law Firms industry page addresses firms on knowledge transfer and repeat drafting, and a Professional Associations page addresses member document portals. Legal adoption is evidenced by name rather than asserted: Holland & Knight, Stewart McKelvey with a published automation journey, Deloitte Legal, LexisNexis, and the Bronx and New York County District Attorneys' offices, the last of which has its own prosecution automation case study. The interface and pricing pages are maintained in English, German, Spanish, French, Czech and Slovak, and three hosting regions serve the Americas, Europe and Asia-Pacific. What holds this at B is that legal is one of seven departments and one of eight industries, with no practice-area breakdown, no jurisdiction-specific legal content and no matter-type coverage statement anywhere. A buyer can see that legal teams use it; they cannot see which legal work it is built for. Checked 7 September 2026.