P
Prime Legal AI
Prime Legal AI is an AI workspace for German lawyers, legal departments, courts and public administration that analyzes whole case files, drafts pleadings and letters, reviews contracts and researches the law. A lawyer creates a case file, uploads documents, emails and expert reports, and works on them in chat, with more than 100 tested prompts for tasks from first client inquiries to final pleadings. Research draws on more than a million decisions, on statutes and official documents at EU, federal and state level, and on publisher material, with the Otto Schmidt civil law module added in the ADVANCED plan.
Sources are shown with each answer, and the language models, which the vendor selects automatically and does not name, are kept separate from a semantic search trained for legal text. An anonymization layer replaces personal data before processing. A Word add in, document transfer from DATEV Anwalt and RA-MICRO, and an API connect it to existing systems. CORE costs 199.90 euros and ADVANCED 299.90 euros per user per month on annual billing, plus VAT.
QNC GmbH of Hannover, founded by the lawyer Michael Friedmann, has developed it since 2017 and is part of anwalt.de services AG.
Capability grades
All 15 axes, graded from public sources on the date shown. 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.
Prime Legal AI's home page calls the product a universal legal AI workspace for case file analysis, drafting and research, and every paid function runs through it. That covers chat over a whole case file, drafting of pleadings and letters, contract review and suggested next steps. Both plans include unlimited queries and uploads, so only the seat is metered. Beyond the models the product carries content, which Prime Legal AI's FAQ says is the only difference between the two plans. That content is sold as material for the models to work on rather than as a database to browse.
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.
The Prime Legal AI FAQ says the sources used appear directly in each answer, that a button gives an overview of all sources consulted, and that results carry citations for traceability. It then says that, as with any language model system, hallucinations can occur, and that a residual risk remains with missing context, unsuitable sources, very complex tasks and imprecise instructions. It distinguishes a hallucination, which invents content, from a professional error, which misapplies correct content.
The home page promises transparent, verifiable output and, with correct use, no hallucinations, so the two pages disagree. No accuracy figure, test set or error rate is published, and the Terms have no clause on the accuracy of output.
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.
The FAQ answer on hallucinations tells users to check all results against their own professional judgment, and the answer on working a matter ends by telling the lawyer to check and edit the proposal against their own legal assessment. The home page describes case file intelligence that actively suggests next steps, and the product works across whole case files rather than single prompts. Sources are shown with each answer.
Nothing describes a mode that runs unattended, a point at which the system stops for approval, or a record that a lawyer reviewed an output. Beyond a help button in the chat for reporting problems, nothing says what happens after an error, and the Terms have no clause on the use or review of AI 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.
The home page quotes lawyers by name and firm, among them Dr. Matthias Schwenke of PwC and Markus Hartung of Chevalier Rechtsanwälte. Oliver Fouquet of kgh reports one to two hours saved a day, and Katharina Larverseder more than 50 percent of time saved, both as their own estimates. A customer section shows names including DATEV, Sixt and NORD LB, and the home page reports more than 2,500 engaged users. No case study with a dated deployment, a measured before and after, or a method behind the time figures is published.
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 Prime Legal AI FAQ says personal data is processed under a data processing agreement and that the vendor is contractually bound to keep lawyers' professional secrets. An anonymization layer recognizes and replaces names, addresses and bank details, among other personal data, before processing, and the DATEV transfer is anonymized too. Section 5 of the Terms binds both sides to keep business secrets and confidential information, for five years after the contract ends and indefinitely for trade secrets.
The data processing agreement itself is not published. The Terms do not mention section 203 of the Criminal Code or section 43e of the Federal Lawyers' Act, and what the model providers keep is not stated.
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.
The product is sold to lawyers, legal departments, courts and public administration, and users are told to finish every proposal on their own legal assessment. Neither the Terms nor the home page or FAQ says the product does not give legal advice or replace a lawyer's assessment, and the prompt library offers a prompt named legal advice. Jurisdiction limits are not stated: nothing says which German states or foreign systems the sources cover.
The About section says the merger with anwalt.de, a platform for acquiring clients, aims to cover the whole client journey in law and set new standards for digital legal advice, without saying where the tool stops and the lawyer begins.
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.
Asked how the product meets the GDPR, the Federal Lawyers' Act and the AI Act, the FAQ answers on data processing and secrecy. On the AI Act it says only that the vendor helps firms build the AI literacy Article 4 requires through training, and it does not classify the product. Quality is attributed to design choices in the search and the prompts rather than to any review body. Prime Legal AI's home page shows a seal of the German research allowance certification body for in house research and development.
No accountable owner for model behavior, release testing, evaluation across practice areas or kinds of matter, or finding about uneven output 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.
Section 12 of the Terms makes the vendor a processor acting only on the firm's instructions, with the firm as controller and responsible for requests from data subjects. No subprocessor list, breach notice period or access control description appears on the home page, the FAQ or the Terms, and retention is left open. The privacy policy covers the website alone, listing analytics, advertising and newsletter services. The footer points to a trust center.
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 10 of the Terms of 22 April 2026 makes the vendor liable for intent and gross negligence by itself or its agents, under product liability law and for personal injury. For slight negligence it is liable only for breach of an essential contractual duty, with property and financial loss limited to damage typical of the contract and foreseeable, including lost profit. Other remote consequential damage is excluded.
Section 10.6 states business liability insurance with annual cover of 2 million euros in total. Section 6.1 warrants the contractually owed quality of the services, with defect claims time barred after one year, and section 10.4 limits liability for lost data to recovery from proper backups. Section 4.7 runs the other way, with the customer indemnifying the vendor against third party claims from unlawful use. No indemnity for the customer, warranty on output content or monetary cap appears, and the Terms have no clause on AI generated results.
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.
The Word add in lets a lawyer check, revise and insert text directly in Microsoft Word, and results export to Word on the firm's own letterhead or to PDF. Documents from a DATEV Anwalt case file reach Prime Legal AI in anonymized form, and the DATEV marketplace lists the DATEV Anwalt classic integration and a partnership since December 2024. Files or whole case files from RA-MICRO go through the Prime Legal AI uploader, which works with the on premise version and not the cloud version.
An API is available, connections to stp.one and stpAdvoware are in development, and there is no Outlook integration, although emails with attachments can be uploaded. Nothing describes write back to DATEV or RA-MICRO, what the API exposes, or how the add in is rolled out and configured across a firm.
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.
Section 12.4 of the Terms guarantees that customer data is stored only in Germany, another member state of the European Union or another state of the European Economic Area. Prime Legal AI's FAQ says the product runs only on secure, isolated EU cloud infrastructure. Asked what EU hosting means for the language models, it answers that computing and data processing happen on servers inside the European Union rather than in the United States.
The home page describes the infrastructure as partitioned. The cloud provider, the country or region of storage, where each model runs, and whether hosting differs between plans are not stated. No private instance or on premise option is described.
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.
The FAQ answer on data security says the solution is certified to ISO 27001, and the home page repeats the certification beside EU cloud hosting. The footer links a trust center for the company hosted on Secfix. No certifying body, certificate number, scope, issue date or audit report is stated on the home page, the FAQ, the Terms, the privacy policy or the DATEV marketplace listing. None of them says whether the certificate is held by QNC GmbH, by anwalt.de services AG or by a hosting provider. No SOC 2, BSI C5 or ISO 42001 attestation is claimed.
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.
Asked which AI models it uses, Prime Legal AI's FAQ says the product works with several leading language models, is built to be independent of any one, and uses a connector that selects and steers the model best suited to each task. Around the models sit a semantic search trained on more than a billion tokens of legal text and a knowledge graph of legal sources. The models are kept apart from the legal content database, and their temperature is set to 0.0 by default.
An FAQ answer on EU hosting mentions models such as GPT or Claude without saying those are the ones in use. No provider, model version, hosting route or commitment to tell customers when a model changes is published.
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.
CORE and ADVANCED are priced per user per month on annual billing on the home page and in the FAQ, and their listed contents differ only by the Otto Schmidt module. Section 2.4 of the Terms caps Otto Schmidt queries under fair use. A custom offer for larger or special requirements is available on request, without a price. The Terms bill at the prices in force at signature, and no monthly rate or implementation fee is published.
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.
Prime Legal AI's home page says the workspace serves law firms of all sizes, solo practitioners, legal departments, courts and public administration, for all areas of law and every phase of work. No segment page describes what courts or public bodies use it for, and no practice area page is published. The FAQ suggests switching on additional sources such as state case law when needed. The stated limits are technical ones, such as the civil law scope of the Otto Schmidt module.
The named users span corporate, tax, insolvency and advisory practices, and none is a court or public body. Nothing says which areas of law or kinds of work the product handles poorly.
2 public documents
The public pages on file for Prime Legal AI, with the recorded signals each one supports and the date it was last read. Open any of them and check the reading against the record.
-
primelegal.ai/agb2 signals
Prompt and Output Retention, Ethical Walls and Matter Segregation
Read Oct 2, 2026
-
primelegal.ai/faq1 signal
Client Data in Training
Read Oct 2, 2026
€199.90 per user/monthEUR, as published, never converted
- Prime Legal AI costs 199.90 euros per person per month for CORE.
- ADVANCED, with Otto Schmidt books and journals, costs 299.90 euros.
- Both are paid for a year at a time and VAT comes on top.
- You can try it free for 31 days without a card.
- Big organizations ask for a custom price.
CORE costs 199.90 euros per user per month on annual billing, plus VAT. It includes unlimited uploads and queries, team collaboration, more than 2.5 million primary sources with publisher content, and tools such as the Word add in and the API. ADVANCED costs 299.90 euros on the same basis and adds the Otto Schmidt civil law module of seven commentaries, nineteen handbooks and six journals. Otto Schmidt queries are capped at 200 per user per month under fair use. A custom offer is available on request, and the free trial runs 31 days without a card.
Implementation: Not published.
Confidentiality and data terms: Not applicable in the German market. The Terms make the vendor a processor under the GDPR, and the data processing agreement itself is not published.
Note: Prices exclude VAT and are billed annually. The DATEV marketplace listing gives a price from 179.91 euros a month, which does not match the current plans. Contracts renew for the same term with one month's notice of termination before the end (section 13.2 of the Terms of 22 April 2026). No monthly rate is published.
Legal Signals
What each signal meansA signal records what public sources say on the date shown. It is not a grade and it is not a recommendation. Where a signal reads Not addressed, it means the index did not locate the material in public sources on that date, which is a statement about disclosure rather than about the product.
Client Data in Training
Can material a lawyer puts into this product be used to train a model?
A public policy or trust page states no training on customer content, with no matching term located in the published agreement.
Training on inputs is ruled out in Prime Legal AI's FAQ and on its home page. The Terms, the only agreement published, do not address training, and the data processing agreement the FAQ refers to is not published.
Prompt and Output Retention
How long does the product keep what a lawyer typed, and can that be set to zero?
Retention is acknowledged in public materials with no stated period.
Section 13.5 of the Terms tells the firm to back up its data before the contract ends, because access will regularly no longer be possible afterwards for data protection reasons. No retention period or deletion procedure for chats, case files or uploads is stated on the home page, the FAQ or the Terms.
Ethical Walls and Matter Segregation
Does retrieval respect the firm’s ethical walls, or can the model read across them?
Segregation is asserted in public materials with no published detail on how it is enforced.
Prime Legal AI's home page describes a separate, isolated workspace for each user, section 2.2 of the Terms ties each login to one named user, and its FAQ says chats belong to a case file and cannot be shared on their own. No permission model for sharing case files within a team, or for walls between matters, is documented.
Third Party Request and Subpoena Notice
If someone subpoenas the vendor for a firm’s data, does the firm hear about it first?
No located term or policy addresses third party requests for customer data.
Requests from courts or authorities for customer data are not addressed in the Terms, the privacy policy, the FAQ or the home page, and no transparency report is published. The confidentiality clause requires written consent for disclosure and names no legal exception.
Primary Law Corpus Provenance
Where does the law in this product come from, and does the vendor have the right to use it?
Sources are identified without stating the license or rights basis.
The FAQ describes more than a million decisions and statutes, regulations and official documents at EU, federal and state level, and names secondary sources: publishers including Deutscher AnwaltVerlag, a database of more than 600,000 online consultations from Frag einen Anwalt, and the Otto Schmidt civil law module in ADVANCED. Prime Legal AI's home page says the content is extended daily. The origin of the decisions and statutes and the license basis for any source are not stated.
Good Law Verification
Does the product tell you when the authority it just cited has been overruled?
The product instructs the reader to check the citation without performing the check.
Beyond a general instruction to check all results, and the sources shown with each answer, no check of whether a cited decision was later reversed or a provision amended is described.
Refusal and Uncertainty Behavior
What does the product do when the answer is not in the corpus?
No located public material addresses what the product does when it cannot ground an answer.
No path is described by which the product declines to answer or flags an unsupported answer. The FAQ's account of when hallucination risk remains is a caveat, not a refusal behavior.
Fabricated Citation Record
Does a public court record exist addressing fabricated or hallucinated legal citations in output from this product?
No court order, opinion or disciplinary record addressing fabricated or hallucinated legal citations produced by this product has been located as of the date shown. This is a statement about the public record on that one subject, not a finding about the product, and this signal is not a litigation history.
The AI Hallucination Cases database maintained by Damien Charlotin records no case involving Prime Legal AI or its operator. No court order, opinion or disciplinary record naming the product is on the public record.
Bar Guidance Alignment
Has the vendor engaged in public with the ethics opinions its buyers are bound by?
Public materials refer to professional responsibility in general terms without naming guidance.
Compliance with the Federal Lawyers' Act is claimed on the home page and in the FAQ, and that general claim is the only professional reference. No bar association guidance, professional code or ethics opinion on AI is named.
Billing and Fee Posture
Does the vendor address what happens to the bill when the work takes an hour instead of six?
Public materials claim time savings without addressing billing or disclosure, and the product sits inside a fee relationship between a lawyer and a client where those savings would change the bill.
Time savings of up to 90 percent are claimed on the home page without a method, alongside lawyers' own estimates of the hours they save. Nothing addresses how AI assisted work is recorded, billed or disclosed to clients.
Outside Counsel Guideline Readiness
Can a firm get this vendor through a client’s AI clause without a bespoke negotiation?
This signal has not been recorded for this vendor yet. It is not a finding either way.
Court Disclosure Support
If a judge’s standing order requires an AI disclosure, can the product produce one?
Some elements of the record are available, short of a document level export.
The sources behind an answer can be listed, and documents export to Word or PDF. No record for each document of the model used and of human verification exists, and no guidance on disclosing AI use to a court is published.