AI:ssociate
AI:ssociate is an Austrian legal and tax AI assistant. It answers questions from a curated base of statutes and supreme court rulings, analyses your own files and drafts pleadings, always citing a clickable source.
What is AI:ssociate?
AI:ssociate is a legal and tax AI assistant built for Austrian law and published by AI:ssociate FlexCo in Vienna. It was founded by the lawyer Dr Philipp Merzo out of the daily practice of the Viennese business law firm Grama Schwaighofer Vondrak, and is developed by lawyers and tax advisers working alongside machine learning engineers. The stated aim is not another general chatbot but a tool that reasons with Austrian statutes, supreme court case law and legal method.
The product bundles three modules into one system. Recherche answers a question asked in natural language with a first assessment and a citable source behind every statement. Analyse takes your own files, Word, PDF or scanned documents, makes them semantically searchable and checks clauses against the law in force. Draft turns a brief into a complete first draft of a pleading or a contract, from an NDA to a share purchase agreement, then exports it to Word or PDF.
What sets the tool apart is its corpus. The vendor describes a proprietary knowledge base of more than a million pages of legal text, combined with a multi-stage extraction and prompting strategy. It covers the central Austrian statutes, European regulations, roughly 160,000 Supreme Court decisions since 1993, more than 150,000 administrative court rulings from the past twenty years, all of the roughly 20,000 constitutional court rulings since 2000, plus case law from the Court of Justice of the European Union and the federal and regional administrative courts, and the finance ministry's guidelines. Doctrine and commentary are deliberately left out: only primary sources are indexed, and answers are confined to the extracted Austrian legal context, which the vendor says makes hallucinations largely impossible. An automated hallucination check runs on each answer.
Language models from OpenAI and Anthropic do the generation, hosted on European servers, and documents can be pseudonymised locally beforehand. Around twenty named Austrian law firms, in-house teams and tax practices appear as customers on the homepage. The vendor is candid about the limit: answers are generated automatically, are never reviewed by staff, and are not legal or tax advice.
What it does
- Ask a legal or tax question in plain language and receive a sourced first assessment
- Open every cited statute or ruling in full text and jump to the official RIS entry
- Upload pleadings, contracts, official decisions or opinions and search them semantically
- Check a contract clause against current legislation and supreme court case law
- Generate a draft pleading or contract, refine it in the editor and export it to Word or PDF
- Pseudonymise a document before any external model processes it
- Work directly inside Microsoft Word and Outlook through the add-ins
When to use AI:ssociate / When not to
A quick filter to help you decide if AI:ssociate is the right fit.
When to use AI:ssociate
- Austrian lawyers in private practice who want a sourced first assessment before the real drafting begins
- Tax advisers, accountants and auditors structuring client queries across income tax, VAT and procedural law
- In-house legal teams triaging business requests and reviewing their own contract portfolio
- Municipal and public-sector administrators handling public-law questions, permits and historic official decisions
- Generalists working outside their usual practice areas who need a trustworthy starting point
When not to use AI:ssociate
- Anyone without legal training: the terms exclude non-qualified users outright
- Consumers and private individuals, since contracts are concluded with businesses only
- Law students and academic staff, who are not yet eligible although an education edition is announced
- Practitioners working outside Austrian and European law, which is the whole scope of the knowledge base
- Teams looking for doctrinal commentary, deliberately excluded in favour of primary sources
How to use AI:ssociate
A typical end-to-end flow, from setup to results.
- Register on the app subdomain; the one-week Premium trial starts automatically, with no technical installation
- Pick a subscription on the website and fill in the form, which feeds your details straight into the licence agreement
- Sign the contract electronically through DocuSign, or ask for a PDF to sign on paper
- Wait for the vendor to confirm that your organisation meets the business-to-business conditions and return the countersigned contract
- Ask your question in full sentences, naming the relevant keywords or statutory provisions explicitly rather than alluding to them
- Open the cited sources to read the original text, follow the link to RIS, or ask for a ruling to be summarised
- Refine with follow-up questions, keeping one topic per conversation and starting a new chat for a new subject
- Upload your own files by button or drag and drop, then choose whether to query a file on its own or alongside a field of law
- Switch pseudonymisation on at upload, review the placeholder text, add any missed terms by double-clicking and confirm
- In Draft, describe the document, answer the scoping questions, generate, adjust in the editor and export to Word or PDF
Pros & Cons
Pros
- Every statement is backed by a citable source you can open and verify, with a direct link to the official RIS database
- Primary sources only, drawn from a curated corpus rather than the open web, with a quantified body of case law
- Genuinely Austrian specialisation, including local legal method rather than a one-size-fits-all model
- Serious confidentiality engineering: pseudonymisation before processing, AES-GCM encryption and a passphrase the vendor cannot read
- European hosting, an Austrian publisher, signed data processing agreements and a published bar association checklist
- Customer data is never used to train language models, stated without hedging
- One free week of the full Premium edition, monthly cancellation for the first six months and sector deals negotiated with professional bodies
Cons
- Strictly Austrian and European scope, of no use for any other national legal system
- Reserved for legally qualified users and businesses only: no consumers, and no students for now
- No doctrine or commentary in the knowledge base, which is a deliberate choice but a real gap for some research
- Hard monthly quotas per tier, from 25 to 250 queries, 2 to 5 drafts and 5 to 125 uploads
- Annual commitment after the first six months, with prices quoted net, per user, and one licence required per person
- No mobile app and no public API; the Draft module was still in beta at the time of collection
- Conversation context only covers the last answer or one explicitly selected, and answers are never reviewed by a human before you see them
Pricing & Plans
There is no permanent free plan. Every new registration includes one free week of the full Premium edition, after which a paid subscription is required. The lowest entry point is EUR 29.00 per user and per month for the Small plan limited to tax law; the equivalent plan covering all fields of law starts at EUR 39.00. All prices are quoted net, per user and per month, excluding statutory VAT.
- EUR 39.00 per user and month - 25 research queries
- 2 drafts and 5 uploads per month
- covering civil
- criminal
- public and tax law
- EUR 69.00 per user and month - 100 research queries
- 3 drafts and 50 uploads per month
- EUR 99.00 per user and month - 250 research queries
- 5 drafts and 125 uploads per month
- EUR 29.00 per user and month - 25 research queries
- 2 drafts and 5 uploads per month
- EUR 49.00 per user and month - 100 research queries
- 3 drafts and 50 uploads per month
- EUR 79.00 per user and month - 250 research queries
- 5 drafts and 125 uploads per month
- EUR 39.00 per month for 5 extra drafts
- EUR 39.00 per month for 15 extra drafts
- EUR 39.00 per month for 30 extra drafts
- Small at EUR 35.10 per month
- a 10 percent discount
- enterprise and multi-licence deals are quoted by the sales team
Data, GDPR & hosting
A consolidated view of how AI:ssociate handles your data.
GDPR overview
GDPR compliance is claimed explicitly: the FAQ answers the question with a plain yes, and data processing agreements are signed with customers. The controller is AI:ssociate FlexCo, Schottengasse 4/26, 1010 Vienna, reachable at office@aissociate.at. Processing runs on servers inside the European Union and personal data can optionally be pseudonymised before any model sees it. The privacy notice, dated July 2026, gives a legal basis under Article 6 for each purpose and lists the full set of data subject rights: access, rectification, erasure, restriction, portability, withdrawal of consent and objection, together with the right to complain to the Austrian data protection authority under Article 77. No data protection officer is named and no Article 27 representative is designated, which is consistent with a controller established in Austria. A signed checklist from the Austrian Bar is published as evidence of professional-secrecy compliance.
Who owns the data?
You keep ownership of what you upload, but the terms grant AI:ssociate a worldwide, royalty-free, perpetual and transferable licence over those documents, expressly limited to feeding them into your own AI-assisted searches. You warrant that you hold the necessary rights and indemnify the vendor against copyright claims. Where you send personal data belonging to third parties, you act as controller and AI:ssociate as processor, working only on your instructions. Uploaded files are visible to your account alone and account sharing is forbidden. Queries, answers, files and generated documents are treated as confidential and are not passed to third parties unless a legal obligation requires it.
Reuse rights
Whatever you produce with the tool is yours to reuse without asking: nothing in the terms limits what you do with an answer, a generated pleading or an exported contract. Two qualifications matter. First, the vendor states plainly that neither your questions nor the contents of your uploaded documents are used to train language models; inputs are kept only long enough to produce the answer and never enter a training set. Second, if you use the feedback function you agree that encrypted content may be decrypted and read by the vendor to improve the service and fix errors. Answers themselves carry a standing caveat: they are automated research results, are not reviewed by any member of staff, may be wrong, and must be checked by a qualified professional before use.
Data retention & training
Hosting summary
Hosting stays within Europe. The vendor states that its web servers stand in Europe and are therefore subject to EU data protection standards, and that the pseudonymisation server runs at Hetzner inside the European Union. The language models come from Anthropic and OpenAI but are operated by Amazon AWS, Microsoft Azure and Google Cloud on servers located in Europe, which the vendor presents as the reason no personal or confidential data reaches providers outside the EU. No specific country is ever named, only Europe and the European Union; the site's public IP address resolves to Hetzner Online in Nuremberg, Germany, but that is a network observation, not a statement by the publisher. All data, whether queries, chat histories or uploaded documents, is stored encrypted with AES-GCM, and the passphrase can be user-defined so that even the vendor cannot decrypt. Transfers to third countries remain possible, based either on an adequacy decision or on the appropriate safeguards of Article 46 GDPR, with details available on request by email.
Things to keep in mind
Risks and trade-offs to weigh before adopting AI:ssociate.
- Answers are never read by a human before you see them and the vendor states they can be wrong: skipping the final professional check would put client work at risk
- Speed can erode judgement over time, since a plausible first assessment arriving in seconds is precisely the kind of output that stops being questioned
- The licence granted over uploaded documents is drafted very broadly, even though its stated purpose is limited to your own searches
- Using the feedback function authorises the vendor to decrypt and read the content concerned
- A forgotten user-defined passphrase makes the encrypted content permanently unrecoverable, with no way back for the vendor either
- The privacy notice lists only categories of recipients, advertising, marketing and tracking providers among them, rather than a named list of subprocessors, and transfers to third countries are detailed only on request by email
- Costs scale with headcount because prices are per user and net of VAT, and the contract turns annual after the first six months
Setup & Integrations
Technical difficulty
Very low. Nothing is installed: the product runs in a browser and the vendor puts the whole onboarding at under five minutes. What actually takes time is contractual rather than technical, since you pick a plan, fill in a form that feeds the licence agreement, sign through DocuSign or on paper, and wait for the vendor to confirm that your organisation meets the business-to-business conditions before the account is unlocked. Each user then creates their own account and the publisher activates everyone attached to the organisation. The only technical setting worth knowing is the encryption passphrase in the profile. The Word and Outlook add-ins install from the Microsoft marketplace.
Deployment
Integrations
Supported languages
Behind AI:ssociate
Social
Resources
All the official URLs gathered for verification and reference.
Alternatives
Tools that compete with or complement AI:ssociate.
Frequently asked questions
How is AI:ssociate different from ChatGPT?
Which areas of law does it cover?
Are my questions or documents used to train AI models?
Where is my data hosted?
Is there a free trial?
What does it cost?
Can I use it as a private individual or a student?
Which languages does it work in?
Are the answers always correct?
Is there a mobile app or a public API?
Should you pick AI:ssociate?
AI:ssociate is a deliberately narrow product, and that is its strength. It serves one legal system, Austria's, and one kind of user, the legally qualified professional, and it declines the temptation to be a general assistant. Verifiability is the pivot of the design rather than a feature bolted on afterwards: every statement carries a citation you can open, read in full and follow through to the official RIS database, and the corpus is limited to primary sources whose volume the vendor quantifies openly.
The confidentiality engineering is unusually concrete for a tool of this size. Documents can be pseudonymised by a local model before anything reaches an external provider, files are encrypted with AES-GCM under a passphrase the vendor need not know, hosting stays inside the European Union, data processing agreements are signed, and a checklist from the Austrian Bar is published with technical documentation. Around twenty named law firms, in-house legal teams, tax practices and public bodies vouch for it on the homepage, which is a stronger signal of real adoption than most directories see.
The reservations are just as clear. This remains an assistive system: answers are produced automatically, are never reviewed by a member of staff, and the vendor says so plainly, so a final professional check is mandatory rather than advisable. The Draft module was still in beta at the time of collection. Monthly quotas are tight at the lower tiers, prices are per user and net of VAT, and the contract turns annual after six months. Outside Austrian and European law the tool has no purpose at all. Read as what it claims to be, a research and drafting instrument for Austrian legal and tax practice, it is a serious and well-documented offer.
- Choosing a selection results in a full page refresh.
- Opens in a new window.