// Blog
Notes on AI-native law firm work.
Posts on professional practice rules, architecture and the actual day-to-day of using AI in a law firm.
Posts on professional practice rules, architecture and the actual day-to-day of using AI in a law firm.
Adoption, cancellation rates, the end of the hype: what Thomson Reuters, Gartner and Forrester say about agentic AI, and what follows from it for law firms.
How I run Law Flow with eight AI agents on Claude Code: an event bus, a LangGraph coordinator on AWS in Frankfurt and approvals that no agent can grant.
RAG splits documents into chunks and pulls out the most similar ones. In file work, the most similar passage is rarely the right one.
A usable AI policy answers four questions: approved systems, permitted data, output review and what happens when something goes wrong. The rest is trimming.
The price per user is the smallest item. Infrastructure, integration, running the thing and the time spent checking the output cost more.
A data processing agreement under Art. 28 GDPR satisfies data protection law, not professional conduct law. What law firms need on top of it.
Why law firms end up running AI tools nobody approved, where the professional-conduct risk actually sits, and what works better than a ban.
Chat interface, API or cloud deployment: the same AI model can be professionally out of the question or a workable basis, depending on how a firm reaches it.
The most dangerous AI output is the plausible one: cleanly formatted and still wrong. Why a bigger model does not fix that, and why structure does.
Why law firms should not introduce AI app by app, but as a layer beneath the whole working day. The thesis behind Law Flow.
On AWS Bedrock the inference profile decides the place of processing, not the model. How law firms bolt the EU boundary shut in technical terms.
Why law firms should start with filing, deadlines and KYC rather than the brief, and how legal-ops-germany puts that into practice.
FIDO2, MFA, a retired root user, PrivateLink, your own KMS keys, cost alerts: six hardening steps before the first client document.
Six architecture patterns for AI agents compared. Why traceability usually points law firms toward sequential workflows.
Everything is a trade-off, says the First Law of Software Architecture. Why the judgment call matters more in the AI era than command of a programming language.
Text form, formal warning, need to know, subcontractors: the four cumulative requirements § 43e BRAO places on a law firm's cloud contract.