In one sentence
You won't recognise the best AI for your firm by its feature list, but by whether it meets seven hard criteria — and whether it passes a two-week test using your own, anonymised files.
Four categories, not one
According to the Future Ready Lawyer report by Wolters Kluwer, 63.6% of German law firms actively use AI — two years ago it was under 30%. "Legal AI" isn't a single product, though; it breaks down into four types of tool:
- Legal research — finds statutes and rulings with a citation.
- Document automation — generates contracts and pleadings from templates.
- Firm co-pilots — work with the case context: deadlines, facts, client communication.
- General-purpose LLMs — generic chatbots with no legal context.
No provider is strong in all four fields. So anyone comparing should first compare categories, not brands.
The seven criteria that matter
- Genuine document understanding — does the tool work on the file you uploaded, or guess from its training?
- Verifiable sources — every legal statement with a citation, checked against real statutory and case texts.
- EU hosting — processing in the EU, GDPR-compliant, no detour through third countries.
- Data control — no training on your client data, a transparent list of sub-processors.
- Honesty about limits — does the tool say "I don't know"? Internationally, around 1,600 AI hallucination cases have been documented across 35 countries; the Kammergericht Berlin rebuked fabricated citations and reaffirmed your duty to verify.
- Lawyer connection — a deadline second-check, an autonomy dial, handover to a firm.
- Price transparency — a comprehensible model, no hidden token costs.
The framework is strict: the professional duty of confidentiality makes any tool that processes client data outside the EU or stores it for training simply unusable. From 2 August 2026, the transparency obligations of the EU AI Act (Art. 50) also apply — AI-generated content must be labelled, with fines of up to €35 million or 7% of global annual turnover.
An honest assessment — by category
- Research tools shine at the citation but don't know your file.
- Automation tools are fast on standard documents, weak on the individual case.
- General-purpose LLMs are cheap, but without an EU guarantee, citation checking and confidentiality-readiness — why that isn't a firm tool.
- Firm co-pilots like SmartLegalPro rely on case context: a fact graph, a deadline second-check, EU hosting, quota instead of euros, a Mandantenportal with per-document approval, and a Rechtsbibliothek that checks citations against real sources.
Assigning specific competitors letter grades would be dubious — performance depends on your subject matter.
Which profile needs what
- Solo lawyer — priority on an immediately usable assessment and fast handover, with minimal admin. A 60-second initial assessment without an account and 10 free credits to try it out lower the barrier to entry.
- Five-lawyer firm — case context and deadlines across the team, plus a Mandantenportal with Seats and "firm pays" AI.
- Specialist boutique — depth in one area of law, uncompromising citation checking and, where needed, a premium model (the Fable add-on with EU pinning and firm opt-in).
How to test in two weeks
Don't compare on demo data, but on your real cases:
- Week 1 — Choose two or three anonymised real files (PII masking before upload) and give the same task to three providers. Check three things: are the sources correct? Is the deadline right? Does processing run in the EU?
- Week 2 — Test the surrounding workflow: handover, Mandantenportal, per-document approval.
Anyone who, after these two weeks, can say which tool held up against their own file has the most honest comparison there is. Start with the Anwaltsportal.
