Elias Reiche

Hello, good to meet you. I am a fully qualified German lawyer and I build AI infrastructure for law firms and legal departments.

After studying at Humboldt-Universität zu Berlin I founded a legal tech startup and served on the board of the legal tech initiative eLegal. I worked as a research assistant at YPOG and completed my legal traineeship in Berlin, including a stage at Freshfields. Alongside that I am studying legal informatics for an LL.M. at the University of Passau. Now I advise on introducing artificial intelligence safely, as infrastructure rather than as a single tool.

Career

1. since June 2026
   Forward Deployed Engineer · Law Flow My own consultancy for AI infrastructure in law firms and legal departments: architecture on AWS Bedrock (Amazon Web Services is Amazon’s cloud platform, Bedrock its service for language models) with enforced EU data residency (the data cannot leave the EU for technical reasons), AI agents (programs that carry out several work steps on their own) for non-billable work, knowledge bases and context layers, and training for the teams.
2. January – September 2025
   Trainee lawyer (Referendar), M&A/Corporate · Freshfields M&A transactions and post-M&A disputes: review of share purchase agreements, claims notices, exhibit lists. Due diligence, in particular chain-of-title analyses (the unbroken chain of previous share transfers). Opinions on stock corporation law, law of the GmbH (the German private limited company) and transformation law.
3. October 2022 – December 2023
   Research assistant · YPOG Drafting tech contracts (SaaS, AIaaS and PaaS: software, AI models and platforms rented as an ongoing service rather than bought; plus licence and data transfer agreements) and data protection compliance for clients in AI, health tech, biotech and blockchain. Due diligence reports on IP/IT and data protection. Talk on "AI and Intellectual Property" at the Young Lawyer's Forum; contributions to talks at GRUR e.V. and Bitkom e.V. on AI regulation and licensing.
4. June 2022 – February 2024
   Founder · FEATRme Co-founded a legal tech startup: a web app (an application that runs in the browser without installation) for drafting, negotiating and legally concluding contracts in under a minute. Piloted in several advertising agencies, partnership with the IP boutique NORDEMANN, programmes at UnternehmerTUM and the Legal Tech Colab. Responsible for product development from the idea to the launch. Exit at the start of the traineeship.
5. 2018 – 2022
   Student assistant · German Bundestag Two placements, with a member of parliament and in a working group: substantive research, constituent enquiries, preparation of committee sessions. Before that, freelance work for the Federal Foreign Office at two international conferences.

eLegal: legal tech as part of legal education

eLegal is a student initiative for legal tech and the future of law, founded in Göttingen in 2019 and active across Germany. It set out to get law students interested in practice-relevant future topics while they are still at university and to create a forum for interdisciplinary exchange. The association now has over 150 members from all over the country (students, trainee lawyers, research assistants and practising lawyers) and is the student partner of the Legal Tech Verband Deutschland.

In 2023 Elias Reiche was the board member for finance and partnerships. That year saw the scaling of Legal Tech University, a free learning platform on legal tech developed together with law firms, startups and universities, which launched in February 2022. The initiative also runs the podcast How to Legal Tech, an interview series, a newsletter, workshops and a legal tech hackathon.

The board role (finance and partnerships) was less legal than structural: budgets for a volunteer team spread across the country, cooperations with law firms and legal tech companies, and the question of how a free educational offering sustains itself. That perspective still shapes the work at Law Flow: digitalisation in the legal market not as a product question, but as a question of incentives, education and organisation.

Note: the association’s former domain elegal.technology is no longer in the hands of the initiative and redirects to an unrelated site. Current information is on eLegal’s LinkedIn profile.

Focus areas

• Professional conduct rules and AI§ 43e BRAO (the professional rule on engaging service providers), § 203 StGB (breach of professional secrets), Art. 28 GDPR and the BRAK guidelines: what a law firm has to secure contractually and technically before client data reaches an AI system.
• AI infrastructure and EU data residencyAWS Bedrock with EU hosting enforced through Service Control Policies (account-wide blocks that not even an administrator can lift), your own KMS keys (encryption keys that sit in the law firm’s account), hardware MFA (sign-in only with a physical security key), PrivateLink (a connection to the service over a private network instead of the open internet) and invocation logging without content capture (a record of who called which model when, without writing down the inputs and answers).
• Knowledge management and agentic searchContext and knowledge layers for law firms: as a structured file tree with retrieval at run time instead of a vector database (a database that makes texts searchable by meaning rather than by keyword), for as long as the volume does not force otherwise.
• Non-billable automationDeadline calculation, filing, incoming post, conflict checks and anti-money-laundering checks: the work where AI can be tried out without putting a matter at risk.
• Deterministic executorsWherever deadlines, figures or money are involved, an executor (a hard-coded calculation module that returns the same result for the same input) does the arithmetic and shows its working, instead of a language model estimating.
• Contract and IT lawSaaS, licence and data transfer agreements, data processing, IP/IT due diligence. From the work at YPOG and in my own legal tech startup.

Principles

Confidentiality is an architecture decision, not a contract clause. If the EU border is not technically locked, no assurance in the data processing agreement will help. What that looks like in practice is in the post on enforced EU hosting.

The starting point belongs in non-billable work. Deadline control, filing and conflict checks are the low-risk place where a firm learns what AI can do and what it cannot, at length here.

Trade-offs remain human work. A language model formulates the weighing up, it does not do it. Why that matters more in the age of AI.

The best engagement ends with the firm no longer needing the advice. Legal tech is not software you buy, it is a capability you build.

Open source

Out of this work comes legal-ops-germany, an open library for the non-billable work of German law firms: deadline calculation under §§ 186–193 BGB and § 222 ZPO, RVG and GKG fee calculation, anti-money-laundering screening, conflict checks and DATEV export, each as deterministic executors (hard-coded calculation modules instead of estimates from a language model) whose calculation stays auditable.

Contact

Email: info@law-flow.de
LinkedIn: linkedin.com/in/eliasreiche
GitHub: github.com/eliasreiche

The formal details required under § 5 DDG are in the legal notice.

Law Flow provides general consulting and technical implementation support, but not legal advice and no legal service within the meaning of the German Legal Services Act (RDG).