LEGALFLY vs Noxtua: how they compare in 2026
LEGALFLY and Noxtua are both European legal AI platforms, one from Ghent and one from Berlin, and both sell sovereignty as the proposition. Noxtua owns its stack. It runs proprietary models trained in house on legal data licensed from national publishers, named individually as C.H.Beck for Germany, Poland, the Czech Republic and Slovakia, MANZ for Austria, Ciela for Bulgaria and Blendow for Sweden. Those models run on European infrastructure it identifies to the facility, including Deutsche Telekom's Industrial AI Cloud in Munich. It also states compliance with the German professional secrecy provisions at section 43e BRAO and section 203 StGB. LEGALFLY answers on architecture rather than ownership. Documents are anonymized before analysis begins, and that component can run inside the customer's own environment so identifiable data never crosses to the tenant. That is one of three deployment models it publishes, with the differences spelled out.
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.
Discovery takes a request in plain language and decides whether to research it, draft from it or route it to a specialist agent, grounding the answer in legal sources and the team's own files. It can also edit an uploaded document or generate Word, Excel and PowerPoint output. Review, Draft, Multi Review and Legal Radar each run models over contracts, data rooms or regulatory feeds, and Agent Studio chains those agents into workflows with approval steps. Legal Radar scans regulators, national banks, official legal databases and courts around the clock, then maps each new obligation to the team's policies and playbooks with an impact summary. Playbooks themselves can be generated from a team's guidelines or golden contracts, with risk levels, thresholds and rules set per jurisdiction. Before any of that work begins, a model pseudonymizes names, roles, company details and identifiers in text, images and handwritten notes.
The AI is the product, built down to the infrastructure. The vendor says it is built as a complete system controlling infrastructure, model, data and interface, running proprietary models trained on licensed legal data rather than calling third party foundation models. It also publishes its own legal embedding model, Noxtua Voyage Embed, built with Voyage AI and dejure.org. Remove the models and nothing remains.
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.
Discovery answers carry citations and reasoning a user can audit, drawn from case law, legislation and the team's own knowledge collections, and the user chooses the sources and jurisdictions behind each answer. Review explains every flagged clause with its reasoning and supporting sources, and Multi Review links each extracted field back to the document it came from. The knowledge base combines more than 700 government and official sources with FromCounsel commentary written by UK barristers and Qistas coverage of nine or more Gulf and Arab jurisdictions, updated daily. Its stated reach is 35 jurisdictions across the EU, UK and Middle East, while product pages cite more than 130 for research and more than 110 for translation. Sources are described as verified, and anonymization is said to keep the full legal meaning of each review. No measured accuracy accompanies those answers.
Grounding is real and documented, with published comparative measurement for one component but no accuracy figures for the product itself. The retrieval method is described in detail. The agent analyzes a query, develops a solution plan and researches systematically across licensed publisher content, and version 5 generates a knowledge graph for each query that surfaces relevant sources and maps how they connect. Output carries source references into a licensed corpus a reader can open. The vendor publishes measured figures for its own retrieval model: Noxtua Voyage Embed outperforms a named OpenAI embedding model on legal text benchmarks, with 1.7 times better search accuracy and 2.2 times better ranking quality at a third of the dimensionality. That measures retrieval, not answer accuracy. Not located: an accuracy or hallucination rate for generated output, a test set for it, or any abstention behavior.
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.
Agent Studio runs whole workflows from intake to approval. A workflow starts from email, Slack, Teams or a manual trigger, conditional logic routes the work and sends follow ups, and the system escalates when set thresholds are met. Teams place approval and review steps wherever a decision needs a person, who is notified when work finishes or needs attention. Elsewhere a lawyer stays in the review seat. Review produces redlines that a user accepts, rejects, edits or applies in bulk as tracked changes, and Draft output passes through review and approval before it is shared. Through the Word and Outlook add ins, sales, procurement and HR staff handle routine contracts inside guardrails that legal sets in the playbook. The terms put the duty to check every output on the customer.
Autonomy is claimed and oversight is asserted without a mechanism. The vendor describes agentic AI that independently develops a solution plan and executes systematic research, and states results are traceable at every stage, which is transparency of process rather than a control structure. Searched the site, the three product pages, the security page and the press releases and located no description of what the system does on its own versus what a lawyer must approve, no review surface, no threshold at which the agent stops, and no statement of what happens after an output is wrong. Traceability shows a reader what happened; it does not establish who decides.
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.
Three customer stories name the company, a spokesperson and a result. ECS, a transport and logistics operator with 650 staff that screens 200 to 250 contracts a year, reports twice the legal capacity and 50% time saved. Adeera, a hotel operator in the PIF group, reports review time cut by 93% from two to four hours per contract, with its general counsel quoted. Wealins of Foyer Group, a Luxembourg life insurer, reports 20% time savings after moving regulatory monitoring onto Legal Radar, with its chief executive quoted. The clients page adds a 2026 contract with the European Commission's HR Directorate General for complex HR legal workflows, and quotes from Gimpel Private Markets, MSIG Europe and Duvel Moortgat. Each figure is the customer's own account.
Customer logos and an unattributed selected clients strip stand in for evidence. A customer stories page exists in the navigation, and a named endorsement appears from Dr Markus Kaulartz, partner at CMS, though CMS is also a co-initiator and investor rather than an arm's length customer, which is disclosed on the page and weakens it as independent evidence. Corpus scale figures are published, 130 million plus searchable documents and 8.5 million plus court decisions, but those describe the database rather than a deployment. Searched the site, the customer stories entry point and the press center and located no named deployment with figures, dates and an assessable method.
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.
Pseudonymization is the center of the confidentiality model. Names, roles, company details and identifiers are replaced before analysis, and a team can adjust the level or whitelist terms that must stay visible. In the on premises option the anonymization component runs inside the customer's environment, so only pseudonymized data reaches the dedicated tenant. Anthropic and OpenAI, the model providers in the trust center, receive pseudonymized content only unless a customer turns that off, under zero retention and no training. Section 5.2 of the terms says customer data will not train the service without the customer's consent, and section 5.1 leaves the customer exclusive ownership of its data. Each client's environment is isolated. Attorney client privilege and work product are not addressed.
The confidentiality posture is complete and engages the professional secrecy question in statute. The vendor states compliance with Section 43e BRAO and Section 203 StGB, the German provisions on what a lawyer bound by professional secrecy may use, and says this permits use by confidentiality bound professionals without anonymization. Training is addressed directly: inputs and outputs are never used to train the AI and never shared with third parties. Separation is documented: every project is encrypted in transit and at rest under its own cryptographic key, access is strictly controlled, and every action is logged. The vendor also says it has no persistent access to plaintext data. Certification covers AI governance as well as security. The customer agreement itself is not published, so these commitments were read from the security page rather than a contract.
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.
Section 10.2 of the terms describes the service as a tool for professionals working with proper care, says it gives no legal guidance or advice, and makes the customer responsible for verifying every output. The navigation sells Discovery as legal research and advice, and the Word and Outlook add ins put routine contracts in the hands of sales, procurement and HR staff within legal's guardrails. Competence and supervision duties are not addressed, and no jurisdiction limit is set beyond the coverage counts.
A real position is published and it is grounded in named law rather than a disclaimer. The vendor addresses the professional rules that govern whether a lawyer may use the tool at all, citing Section 43e BRAO and Section 203 StGB and alignment with the EU AI Act, and states the audience precisely across law firms, in house teams, auditors and tax advisers, courts and public authorities. Jurisdiction limits are handled structurally rather than in prose: the product ships as jurisdiction specific editions and the site requires a jurisdiction selection before granting access, which is a real published boundary. Competence and supervision duties are not addressed, and the position is framed around the vendor's own permissibility rather than around what the lawyer remains responsible for.
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.
Governance shows up as product controls. Agent Studio and the platform pages describe governed outputs through playbook enforcement, approval steps and role based access, and auditable actions with transparent reasoning. Its trust center controls include an information security policy, a risk assessment and treatment policy and a vendor risk management policy. A report on the AI governance gap sits among the product resources. The controls govern who can act on output. How the models themselves are tested, and who answers for their behavior, is not described.
A published governance framework with real substance and independent validation, short of testing results or a named owner. ISO 42001 certification covers the AI management system specifically and is stated alongside ISO 27001, 27017, 27018 and 9001, BSI C5 and TISAX, with the vendor stating controls are independently and regularly audited and publishing a trust center for the full current list. Alignment with the EU AI Act is stated. That is a governance regime for AI rather than a principles page, and it is externally audited. Not located: a named internal owner of model governance, published pre release testing results for model behavior, or any disclosure about uneven output across matter types, parties or populations.
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.
Documents are encrypted at rest and in transit, and every client runs in an isolated environment. Access is set through attribute based access control, multifactor authentication and session governance, with single sign on through Okta, Azure AD or Google Workspace over SAML 2.0, OIDC or OAuth. An outside firm runs penetration tests every year. The trust center, monitored through Secureframe, lists five subprocessors: Microsoft Azure for infrastructure, Auth0 for identity, PostHog for product analytics, and Anthropic and OpenAI for AI inference. It names an incident response plan, a data retention and disposal policy and business continuity controls, without their text. The privacy policy keeps pseudonymized contract content to improve the service, and keeps personal data as long as its purpose requires.
Substantive published policy covering most of the ground. Per project encryption in transit and at rest with a distinct cryptographic key per project, strictly controlled access, full action logging, a stated position that the vendor has no persistent access to plaintext data, regular security reviews and risk assessments, and a Vanta backed trust center at a stable URL. Infrastructure providers are named individually with their own certifications, which is effectively a partial subprocessor disclosure at the hosting layer. Not located: a stated retention period or deletion control for documents and prompts, a full subprocessor list, and an incident or breach notification practice.
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.
Section 8 of the terms caps LegalFly's total liability at the fees paid in the 12 months before the claim, and the cap also covers its indemnity. Indirect, consequential and punitive damages are excluded, and LegalFly is not responsible for losses from decisions made on generated content. Against third party claims that the customer's use of the service infringes intellectual property, LegalFly indemnifies the customer. It can secure a license or modify the service, and pays damages from a final judgment or an approved settlement. The indemnity excludes claims arising from customer breach, outside modifications, use outside the documentation, configurations the customer specified and combinations with unapproved products. The warranty promises professional service with reasonable skill and care. Warranties on the accuracy, completeness and reliability of output are disclaimed, and the customer verifies all output. The trust center offers cyber, liability and fraud insurance certificates on request. The terms of 24 April 2025 fall under Belgian law and the courts of Ghent.
No published indemnity, liability cap, carve out, warranty on output or insurance position is published, and no customer terms or general conditions are published at all. Access to the product itself runs through a jurisdiction gated request form, so the agreement is reached through a sales process. Notable given how much else this vendor publishes: the compliance posture is documented in statutory detail while the allocation of loss when output is wrong is not addressed anywhere public.
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.
Microsoft Word and Outlook carry most of the integration work. In Word a user reviews, redlines, drafts and compares contracts with tracked changes and formatting kept, applies playbooks and turns any contract into a template. In Outlook a user asks about a thread or attachment, drafts replies and fills templates from the details in an email. SharePoint and Google Drive connect so documents can be imported or read by Discovery where they sit, and Multi Review pulls whole data rooms from SharePoint or a document system. Agent Studio workflows start from email, Slack or Teams. Every connection runs anonymization first. No connector for iManage or NetDocuments is named.
Integrations are named without documentation an implementer could use. A Microsoft Word integration is stated on the drafting product page, with work possible either in Noxtua directly or in Word. Searched the site, the three product pages, the security page and the FAQ and located no other integration: no document management connector such as iManage or NetDocuments, no Outlook, no contract lifecycle or matter management, and no integrations page at all. Nothing describes what the Word integration moves, in which direction, or what an administrator configures.
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.
Three deployment models are offered. SaaS is fully managed and monitored. Single tenant gives a customer a dedicated, isolated instance in its preferred Azure region, run by LegalFly. Single tenant with on premises anonymization adds an anonymization component inside the customer's environment, so only pseudonymized data travels to the dedicated tenant over private, encrypted connections, and identifiable data never leaves the customer. The security FAQ adds private cloud, hybrid and full on premises deployment. Residency follows the Azure region the customer picks. The regions on offer are not listed.
Where the software runs and where the data sits is published in concrete detail, because it is the product's whole proposition. Hosting providers are named with their certifications and roles. IONOS is a BSI C5 certified critical infrastructure provider operating the Bundescloud. Deutsche Telekom's Industrial AI Cloud is the AI Factory in Munich behind the Deutschland Stack for regulated sectors including the judiciary. T Cloud Public is BSI C5 certified for critical infrastructures, and T Cloud Public in Switzerland is BSI C5 certified and FINMA compliant, offering Swiss data sovereignty. The vendor says data stays in Europe on infrastructure run by European providers independent of US cloud providers and shielded from the US CLOUD Act, which addresses the jurisdiction of processing and not only where data is stored. What changes between deployment options is not stated.
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.
ISO 27001:2022 and SOC 2 Type II certification are stated, with annual independent penetration testing and a 2026 test listed. The trust center, run on Secureframe, releases the SOC 2 Type II report, a security whitepaper, the information security policy, the ISO certificate, insurance certificates and an architecture diagram through a request form. It shows monitored controls across access, network security, confidentiality, availability, vulnerability management, change management, incident response and risk. The security page describes the framework as both aligned with and certified to ISO 27001 and SOC 2 Type II. Report dates and the auditor are not shown.
The certification set is broad, and a trust center is reachable without a sales call. Named and current: BSI C5, TISAX, ISO 42001, ISO 27001, ISO 27018, ISO 27017 and ISO 9001, with the vendor saying controls are independently and regularly audited. The trust center is a Vanta portal at a stable URL with the full current certification list and policies, a self serve route. BSI C5 and TISAX matter here: they are the German federal cloud computing criteria and the automotive industry information security assessment. No coverage period, report date or named auditor is published on the public pages.
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.
Two AI inference providers are named in the trust center, Anthropic and OpenAI. Each processes pseudonymized content only, unless a customer enables processing without anonymization, under zero retention and no training. Microsoft Azure hosts the platform, and the single tenant option sits in the customer's preferred Azure region. The anonymization step runs before any provider call, and in the on premises option it runs on the customer's side. Which model handles which task is not stated, and LegalFly makes no commitment to tell customers when its providers change.
The vendor owns the supply chain and says so. Models are proprietary and trained in house on licensed legal data rather than sourced from a third party foundation model provider, which is the point of the sovereignty positioning. Where the models run is named down to the provider and facility, including IONOS and Deutsche Telekom's Industrial AI Cloud in Munich. One external component is disclosed with its partner: the retrieval model Noxtua Voyage Embed, built with Voyage AI and dejure.org, with its model identifier published. A customer inherits a dependency chain that is short and named. Not located: a commitment to notify customers when the model or its hosting changes.
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.
Buying starts with a demo, and no price is listed. The terms set out how the subscription works. Each order form caps the number of end users and sets the fees for that number, priced and payable in euros and invoiced annually in advance. The first term is one year and renews automatically, and a customer can lower its user count at renewal with 60 days notice. Fees can rise in two ways, through indexation each 1 January and through an increase at renewal with 30 days written notice, after which the customer has 15 days to terminate. Users beyond the cap can be charged, and professional services such as implementation and training are quoted on a separate order form with estimated fees.
No pricing page exists, no rate is published, no unit of charge is stated and no tier structure appears. The only commercial entry point is a Get Access request that first requires selecting a jurisdiction, so a buyer cannot reach a figure or even a product edition without entering a sales process. Third party pricing figures are not published either.
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.
The product is built for in house legal and business teams in regulated industries. Industry pages cover banks, insurance, manufacturing, technology and professional services, where it pitches due diligence, regulatory research, drafting, contract review and client advisory work. Function pages cover legal, claims, compliance, HR, procurement and sales, each with its own workflows. Coverage is set by jurisdiction. The knowledge base spans 35 jurisdictions of civil, commercial and regulatory law in the EU, UK and Middle East, Legal Radar monitors more than 70, and research reaches more than 130. Named customers range from ECS and Adeera to Wealins, MSIG Europe and the European Commission.
Segment coverage is described with substance and precision. Four segments have their own published sections: in house legal teams, law firms, auditors and tax advisers, and courts and public authorities, the last an explicit positioning for the judiciary and public administration. Jurisdictional coverage is stated as named editions for each country, spanning Germany, Austria, Switzerland, Poland, Czech Republic, Slovakia, Bulgaria and Sweden, each tied to its national legal publisher. Practice areas are not listed beyond a general research, analysis and drafting framing, and nothing says what is not supported.
The 12 legal signals, side by side
Recorded rather than graded. These are the questions a practitioner has to answer before a tool touches a client matter, and the answers are taken from public material only.
Client Data in Training
Can material a lawyer puts into this product be used to train a model?
Section 5.2 of the terms says customer data will not train the service without the customer's consent. The trust center adds that customer data never trains LegalFly's models, and that Anthropic and OpenAI handle pseudonymized content under zero retention and no training. The privacy policy keeps pseudonymized copies of uploaded contracts to improve the service and provide insights.
The security page states directly that inputs and the product's outputs are never used to train the AI and are never shared with third parties, and separately that the vendor has no persistent access to plaintext data. The commitment as located sits on a public security page rather than in a customer agreement, and no customer terms are published on this property, so it is a policy statement rather than a contract term.
The training relationship running the other way is disclosed as well: the vendor states it trains its proprietary models on legal data licensed from publishers, which is publisher content rather than customer content.
Prompt and Output Retention
How long does the product keep what a lawyer typed, and can that be set to zero?
Uploaded contracts are pseudonymized before storage, and the privacy policy keeps that content to improve the service, with no period given. Personal data is kept as long as its purpose requires or the law demands. The two model providers keep nothing under their zero retention terms, and a data retention and disposal policy appears among the trust center controls without its text.
The security page, the data privacy statement, the FAQ, the imprint and the trust center entry point were searched on 29 Aug 2026. Nothing public says how long inputs, outputs or uploaded documents are kept, whether the customer controls the window, or whether deletion is available. The vendor does publish related architecture, encryption for each project under a distinct key and a statement that it has no persistent access to plaintext data, but neither answers the retention question.
Ethical Walls and Matter Segregation
Does retrieval respect the firm’s ethical walls, or can the model read across them?
Access inside a customer runs on LegalFly's own controls. Administrators set attribute based access rules, multifactor authentication and session governance, single sign on enforces the customer's identity policies, and playbooks can be shared with chosen members or teams. Each client's environment is isolated, and the single tenant option separates a customer completely. Walls between matters inside one customer are not described.
The product keeps its own documented separation model rather than inheriting one from a document management system. Every project is encrypted in transit and at rest under its own cryptographic key, access is strictly controlled, every action is logged, and the vendor says it has no persistent access to plaintext data. Cryptographic separation by project is a strong, documented mechanism, but it is the vendor's own model, which the firm must keep aligned with its walls.
No document management integration was located whose permissions retrieval could enforce at query time, and nothing addresses conflicts or ethical walls as such.
Third Party Request and Subpoena Notice
If someone subpoenas the vendor for a firm’s data, does the firm hear about it first?
Section 9.3 of the terms lets either party disclose confidential information when compelled by law, with prior notice to the other party where the law permits. The privacy policy says personal data may go to competent authorities when legally required. No transparency report is offered.
Searched the security page, the data privacy statement, the imprint, the FAQ and the trust center entry point on 29 Aug 2026. No clause committing to notify a customer of a government or law enforcement request for their data was located, and no transparency report was located. The vendor does address the adjacent question of foreign jurisdiction directly, stating data stays in Europe shielded from the US CLOUD Act on infrastructure operated by European providers, which speaks to which state could compel production rather than to whether the customer would be told.
Primary Law Corpus Provenance
Where does the law in this product come from, and does the vendor have the right to use it?
The knowledge base joins more than 700 government and official sources for case law, regulation and legislation with two named publishers. FromCounsel supplies commentary written by UK barristers, and Qistas covers legislation and rulings across nine or more Gulf and Arab jurisdictions. Publisher content and datasets update daily, and teams add their own knowledge collections. The terms for the publisher content are not stated.
Sources are named as national legal publishers, each tied to the jurisdiction it serves: C.H.Beck for Germany, Poland, Czech Republic and Slovakia, MANZ for Austria, Ciela for Bulgaria and Blendow for Sweden, plus a Swiss edition, with dejure.org named in the retrieval model partnership. The rights basis is stated as exclusive publisher partnerships. Scale is given as more than 130 million searchable documents and more than 8.5 million court decisions. No update schedule or lag for this content was located as of 29 Aug 2026.
Good Law Verification
Does the product tell you when the authority it just cited has been overruled?
Answers come with citations and reasoning a user can audit, and legal datasets update daily. Whether a returned case carries a treatment flag or a note of later history is not described.
The site, the three product pages, the jurisdictions page and the press releases were searched on 29 Aug 2026. Nothing addresses whether returned authority carries a treatment signal or whether subsequent history is checked. The Understanding product is described as assessing the validity of clauses, which is contract clause analysis rather than the standing of cited authority. Continental civil law jurisdictions do not use citators in the Anglo American sense, so the question reads differently for a vendor selling into Germany, Austria and Switzerland than for a US product.
Refusal and Uncertainty Behavior
What does the product do when the answer is not in the corpus?
What Discovery does when its sources hold no answer is not described. Agent Studio hands work to a person when set thresholds are met, and Review gives the reasoning behind each flag it raises.
Searched the site, the three product pages, the security page and the press releases on 29 Aug 2026. No published material describes what the product does when it cannot ground an answer, and no explicit no answer path or confidence signal exposed to the user was located. The vendor publishes that results are traceable at every stage and that version 5 surfaces a knowledge graph of the sources behind an answer, which lets a reader inspect the basis of an answer that was given rather than telling them when the system found nothing.
Fabricated Citation Record
Does a public court record exist addressing fabricated or hallucinated legal citations in output from this product?
The AI Hallucination Cases database maintained by Damien Charlotin, which records court decisions worldwide that address hallucinated AI content and the tool involved where known, has no entry naming LegalFly. Its products serve in house legal and business teams in European and Middle Eastern enterprises.
No court order, opinion or disciplinary record naming this product has been located as of 29 Aug 2026. Instrument searched: the AI Hallucination Cases database maintained by Damien Charlotin, which tracks court decisions worldwide involving AI generated hallucinated content and records the AI tool implicated where it is known. Also checked published 2026 sanctions summaries and secondary sanctions trackers. The entries located name filers, and in some rows other products, rather than this one.
This is a statement about the public record on the date shown and not a clearance. The bound is worth stating plainly for this vendor: that database is weighted toward US and other common law filings, and this product sells into German speaking and central European jurisdictions whose decisions are less comprehensively covered by it.
Bar Guidance Alignment
Has the vendor engaged in public with the ethics opinions its buyers are bound by?
Compliance material covers information security and data protection frameworks across the EU, UK and Middle East. No public material engages a bar opinion, law society guidance or ABA Formal Opinion 512.
Public materials engage with named professional obligations by statutory citation: Section 43e of the German Federal Code for Lawyers and Section 203 of the German Criminal Code, the provisions on what a lawyer bound by professional secrecy may use, plus stated alignment with the EU AI Act. The vendor says this permits use by confidentiality bound professionals without anonymization, a specific claim about the rules its buyers follow.
The engagement covers German law only, although the product ships in eight national editions. No equivalent Austrian, Swiss, Polish, Czech, Slovak, Bulgarian or Swedish provision was addressed in located material.
Billing and Fee Posture
Does the vendor address what happens to the bill when the work takes an hour instead of six?
LegalFly sells to legal, compliance, claims, HR, procurement and sales teams in regulated enterprises, and to professional services firms for due diligence, regulatory research, drafting, contract review and client advisory work. Its material is framed around speed and cost. The Wealins customer story reports one legal professional now covering several jurisdictions where coverage was once one or two countries per person, and lower external legal spend.
No per matter record of AI assisted work for fee purposes and no guidance on billing, fees or client disclosure is published.
The site, the three product pages, the customer stories entry point and the press center were searched on 29 Aug 2026. No guidance on billing, fees or client disclosure was located, and no per matter record of AI assisted work for fee purposes. No time savings or efficiency figures were located either. The vendor's framing is precision, verifiability and permissibility rather than hours saved, so there is no savings claim to weigh against the client's side of the bill.
Outside Counsel Guideline Readiness
Can a firm get this vendor through a client’s AI clause without a bespoke negotiation?
Five subprocessors appear in the trust center with their roles, Microsoft Azure, Auth0, PostHog, Anthropic and OpenAI, and the two model providers see pseudonymized content only, under zero retention and no training. The SOC 2 Type II report, security whitepaper and policies are released through a request form. No client facing disclosure template is offered.
A trust center is published at a stable URL, reachable without a sales conversation, carrying what the vendor describes as the full list of current certifications and policies. The certification set named on the public security page is broad and specific: BSI C5, TISAX, ISO 42001, 27001, 27018, 27017 and 9001. Infrastructure providers are named individually with their own certifications, which covers the hosting layer of the supply chain, and the model layer is proprietary rather than third party, which removes the model provider question a firm would normally have to answer for its client. No formal subprocessor list and no consent or notification material for clients were located as of 29 Aug 2026.
Court Disclosure Support
If a judge’s standing order requires an AI disclosure, can the product produce one?
Multi Review exports summaries, comparison tables and diligence packs, with links from each extracted field to its source. Review gives each flagged clause its reasoning and supporting sources and keeps an audit trail, and Discovery answers carry citations. The model behind an output does not appear in these exports.
Several elements of a disclosure record are available and documented. The vendor says every action is logged and agentic research is traceable at every stage, and version 5 generates a knowledge graph for each query that surfaces the sources relied on and maps how they connect, with source references throughout. Sources retrieved and the reasoning path can therefore be recovered. Two elements are missing: no export for each document covering model used, sources retrieved and human verification together was located, and human verification is not recorded as such. The model question is simpler here, since the model is the vendor's own system rather than a changing third party.
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.
- Good Law Verification
- Refusal and Uncertainty Behavior
Which one fits
Choose LEGALFLY if
- Sensitive material cannot leave your environment in identifiable form. LEGALFLY publishes three deployment models with the differences spelled out: fully managed SaaS, a single tenant instance in the customer's preferred Azure region fully segregated from other customers, and single tenant with the anonymization component running on premise so only anonymized data crosses to the dedicated tenant over private encrypted connections.
- You want data minimization applied before the model sees anything. LEGALFLY anonymizes documents before analysis begins rather than after, with adjustable anonymization levels and term whitelisting, environments isolated per client, and the customer retaining data ownership throughout.
- Your requests come from procurement and HR as much as from legal. LEGALFLY publishes dedicated pages for six functions covering legal, claims, compliance, HR, procurement and sales and for five industries, and states plainly that it is built for enterprise legal and business teams rather than for law firm productivity workflows.
Choose Noxtua if
- You need to know exactly whose law the answer rests on. Noxtua names its sources individually and states the rights basis, licensing content from national legal publishers under exclusive partnerships: C.H.Beck for Germany, Poland, the Czech Republic and Slovakia, MANZ for Austria, Ciela for Bulgaria and Blendow for Sweden, across a corpus it puts at more than 130 million documents and 8.5 million court decisions.
- Sovereignty has to mean a named facility rather than a region setting. Noxtua runs proprietary models on European infrastructure it identifies by provider, including IONOS and Deutsche Telekom's Industrial AI Cloud in Munich, states that data stays in Europe shielded from the United States CLOUD Act, and publishes BSI C5, TISAX, ISO 42001, ISO 27001, 27017, 27018 and 9001 through a trust center reachable without a sales call.
- Your obligation is professional secrecy rather than privilege. Noxtua addresses it in statute, citing section 43e of the German Federal Code for Lawyers and section 203 of the Criminal Code and stating that use by confidentiality bound professionals is permitted without requiring anonymization, alongside a statement that inputs and outputs are never used to train its AI and never shared with third parties, with each project encrypted under its own key.
In summary
LEGALFLY
LEGALFLY is a European legal AI platform for enterprise in house teams, positioned as a legal operating system that reaches beyond legal into claims, compliance, HR, procurement and sales. Its products cover research and advice, contract review, drafting, due diligence across data rooms, regulatory monitoring and custom agent building. According to the AI Legal Index, its signature control is anonymization applied before analysis begins, with an option to run that component inside the customer's own environment so identifiable data never reaches a model. It publishes three deployment models and dedicated pages for six business functions. It publishes no training position, no retention period and no price.
Noxtua
Noxtua is a European sovereign legal AI for law firms, in house teams, auditors and tax advisers, courts and public authorities. It ships research, matrix analysis across document sets and drafting with a Microsoft Word add in, as jurisdiction specific editions built with national legal publishers. According to the AI Legal Index, Noxtua's case rests on owning its stack: it runs proprietary models on named European infrastructure, publishes certifications including BSI C5, TISAX and ISO 42001, and states compliance with the German professional secrecy provisions at section 43e BRAO and section 203 StGB. It publishes no customer terms, no liability position and no pricing.
Questions buyers ask
LEGALFLY vs Noxtua: which is better for a European legal team?
Each sells sovereignty a different way. Noxtua owns its stack, running proprietary models on named European infrastructure over content licensed from national publishers, and addresses professional secrecy in statute. LEGALFLY answers with architecture, anonymizing documents before analysis and offering an on premise anonymization option. Neither publishes a price. From the AI Legal Index, based on each vendor's own published materials as of September 3, 2026. No vendor pays for placement.
Where does each one run?
Noxtua names the facilities: proprietary models on IONOS and on Deutsche Telekom's Industrial AI Cloud in Munich, with a Swiss option on T Cloud Public, stating that data stays in Europe on infrastructure independent of United States cloud providers and shielded from the CLOUD Act. LEGALFLY publishes a single tenant instance in the customer's preferred Azure region, and in its third configuration the anonymization layer runs inside the customer's own environment so identifiable data never reaches the tenant. From the AI Legal Index, based on each vendor's own published materials as of September 3, 2026. No vendor pays for placement.
Do either say whether client documents train their models?
Noxtua does. Its security page states that inputs and outputs are never used to train its AI and are never shared with third parties, and it separately discloses that its proprietary models are trained on legal data licensed from publishers rather than on customer content. LEGALFLY publishes nothing either way, which is conspicuous for a vendor whose positioning is built on data protection. From the AI Legal Index, based on each vendor's own published materials as of September 3, 2026. No vendor pays for placement.
Which one names the legal sources behind its research?
Noxtua, in detail. It names the publisher behind each jurisdiction and states the rights basis as exclusive licensing partnerships, shipping as national editions branded jointly with C.H.Beck in Germany, MANZ in Austria and Blendow in Sweden. LEGALFLY states that answers are grounded in verified case law and legislation across its stated jurisdictions without naming a source, publisher, license basis or update cadence anywhere. From the AI Legal Index, based on each vendor's own published materials as of September 3, 2026. No vendor pays for placement.
What do LEGALFLY and Noxtua both leave unpublished?
Neither publishes a price, a tier structure or a unit of charge. Neither publishes an indemnity running to the customer, a warranty on output or an insurance position. Neither states a retention period or a deletion control for documents, prompts or outputs. Neither documents what the product does when it cannot ground an answer. And neither publishes an accuracy or hallucination rate for generated output, although Noxtua does publish comparative figures for its own retrieval model. From the AI Legal Index, based on each vendor's own published materials as of September 3, 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.
Each side leaves one notable gap. LEGALFLY sells data protection as its central proposition, yet publishes no statement on whether customer content may be used to train models. That is silence rather than a commitment in either direction. Its home page also gives two different jurisdiction counts, 110 plus in one place and 130 plus in another. Noxtua publishes no customer terms of any kind, so no indemnity, cap or warranty is readable, and access to the product runs through a jurisdiction gated request form. Its engagement with professional rules also covers German law only, though the product ships in eight national editions. Neither vendor reviewed this page.