The usual break: every chat starts from zero
With most AI tools, every conversation is an island. You open a new chat, re-upload the same documents, and wait for the model to "read" them — and the moment you close the window, that knowledge is gone again. Come back a week later, and the same routine starts over. For a case file with a dozen documents and a lifespan of several months, that adds up to repeated uploads, repeated waiting — and repeated costs.
With us, the knowledge stays attached to the case
We take the opposite approach. You upload documents once into the case file's document management system and let the facts graph build — usually automatically, in the background. The result then stays attached to the case file for good: structured facts, the relationships between them, and an AI summary for each document and for the case as a whole. You can retrieve all of it without the AI ever having to read the file again.
A case file you can click through
Come back months later, and you don't see an empty chat window — you see a case file that knows its own history. Click through the relationship graph: who is connected to whom, which statement contradicts which other one, which piece of evidence supports which claim. Open a single document, and the facts extracted from it sit right beside it, with the exact source. No guessing what actually happened in the case.
Mentally removing a document
A practical example: you accidentally uploaded the wrong exhibit, or a client sends in a corrected version. The obvious question — "What would the facts look like without the reply brief?" — the case file answers directly: deselect a document, and the view shows which facts only entered the case with it, and which earlier statements would apply again without it. This is a pure evaluation of already-captured data — no new AI run, no quota used, no waiting.
And the same case file at an earlier point in time
The same foundation enables a second perspective: time travel. "How did the case file look when the complaint was filed?" — regardless of which document was responsible for a given fact. The time axis and the document axis can also be combined.
Cheaper, because it's reused instead of repeated
Reuse isn't just more convenient — it's also cheaper. Every AI run that draws on already-captured facts instead of the full document text is one fewer paid, repeated call.
New: Lex & Lexi now automatically knows the case file, too
Lex & Lexi, our AI debate pair for pro/con discussions, brief checks, and hearing coaching, previously relied on whatever you fed it within that particular session. A case-linked Lex & Lexi session now automatically pulls in four sources of case knowledge, without you having to re-enter anything:
- the case's facts graph — the same verified, source-linked facts used in the AI assistant,
- your firm playbook, if one is set up — your own standard positions flow into the argumentation,
- the deadline register — Lex & Lexi knows the upcoming dates instead of inventing or overlooking them,
- the next-steps checklist — the case file's current working status.
A debate with Lex & Lexi about a specific case now feels like one with a colleague who already knows the matter — not like an introduction you have to repeat at the start of every session.
Honestly speaking
All of this pays off most for case files with several documents and some duration — that's where reuse genuinely earns its keep. For a single quick question unrelated to a case file, a direct chat remains the fastest route.
→ For lawyers: /fuer-anwaelte · Start verification: /lawyer/onboarding
