Partner at gunnercooke. German-qualified lawyer admitted to the German bar in 2012 (Volljurist), trained at Hengeler Mueller, Freshfields Bruckhaus Deringer, and Cleary Gottlieb. I advise crypto, capital markets, payments and legal AI companies on EU financial regulation, including MiCAR, MiFID II, PSD2, DORA and tokenised capital markets.

I also build software. Before private practice, I worked as a Data Scientist on Python NLP projects at justify.de and Dudenverlag. Today I use Next.js, TypeScript, Python and retrieval pipelines to turn legal workflows into structured products.

For legal AI companies such as Harvey, this is the overlap I care about: I understand the law firm buyer, the regulated client and the engineering stack well enough to translate between lawyers, product teams and commercial teams. I am the kind of European law firm user these products need to convince, and I can also help build the workflows behind them.

• Advise stablecoin issuers, exchanges, payments companies, tokenisation platforms and banks building digital asset desks
• Translate MiCAR, MiFID II, PSD2, DORA and AML/TFR requirements into licensing strategy, product decisions, contracts and compliance workflows
• Work with C-level, product, compliance and engineering teams on regulated business models
• Build legal AI and workflow tools with modern web, data and retrieval infrastructure
• Recruit lateral partners for gunnercooke Germany and design repeatable outreach workflows

Most of my active commits sit in private client, firm and internal product repositories. The public projects below are selected showcases.

I advise on:

• BaFin and ESMA licensing under MiCAR, including CASP authorisations and stablecoin issuer regimes
• Tokenised securities and RWA structures, including securities prospectus drafting and review
• MiFID II and PSD2 perimeter analysis for payments and digital asset platforms
• DORA implementation for regulated entities
• Regulatory correspondence with German and Austrian financial supervisors, including BaFin and FMA

I also publish on the economics of regulated legal services, fixed-fee legal products and the shift from billable-hour work to AI-supported legal outcomes.

• Frontend: Next.js 16 App Router, React 19, Tailwind CSS 4, TypeScript
• Backend: Python 3.12, FastAPI, Pydantic, Node
• Data: PostgreSQL with Prisma, LanceDB for vector retrieval, Neo4j for graph traversal
• AI: Gemini, Vertex AI, Anthropic
• Orchestration: Inngest, AsyncIO, Docker, GitHub Actions
• Deployment awareness: Kubernetes and Cloudflare-style edge delivery patterns

Same commercial product surface Harvey runs on: Next.js, Tailwind, TypeScript and Python, combined with retrieval infrastructure, async workflows and production-grade UX.

A deterministic first-pass linter for MiCAR crypto-asset white paper drafts. Reads a JSON draft, detects the regime (Annex I / II / III), and emits a structured report with pinpoint citations to Regulation (EU) 2023/1114.

Problem: MiCAR Art. 6, 19 and 51 each prescribe a different disclosure regime tied to its own Annex. Drafting reviews catch the same structural gaps over and over — missing reserve composition under Art. 36, missing redemption procedure under Art. 39, missing safeguarding statement under Art. 54 — and doing that read-through manually on every draft is wasteful and inconsistent across reviewers.

What I built: a Python package that applies the right rule set per whitepaper type, ranks findings by severity tied to notification gating, emits text or JSON, and ships with one fixture per regime plus committed sample reports. Zero runtime dependencies, CI on Python 3.11 and 3.12.

Run it:

git clone https://github.com/sebastianforste/micar-whitepaper-linter.git
cd micar-whitepaper-linter
python3 -m micar_linter examples/art-stablecoin.json
python3 -m micar_linter examples/incomplete.json --strict

Repository: github.com/sebastianforste/micar-whitepaper-linter

Internal content engine for regulatory thought leadership grounded in primary sources.

Problem: legal and regulatory content becomes unreliable when drafts are generated without source control, statutory grounding and editorial review.

What I built: a multi-stage editorial pipeline that ingests BaFin guidance, ESMA material, MiCAR delegated acts and recent court rulings, then runs statutory grounding checks before any draft reaches a human reviewer.

Output: LinkedIn posts, carousels and articles under my partner brand.

Stack: Next.js 16, Prisma, PostgreSQL, LanceDB, Neo4j, Inngest, Gemini.

Status: active internal product.

Async pipeline for lateral partner recruiting at gunnercooke Germany.

Problem: senior lateral recruiting requires research, profile structuring, business-case estimation and highly tailored outreach.

What I built: specialised agents run concurrently. A sourcing agent collects public profile information, a profiling agent structures the candidate sheet, an estimation agent models likely book of business by practice area and seniority, and an outreach agent drafts a first message.

Stack: Python 3.12, AsyncIO, Pydantic, Gemini, structured outputs.

Status: used in the gunnercooke Germany lateral pipeline.

Consumer-facing legal workflow prototype.

Problem: many consumer legal issues fail at intake because the facts are unstructured, documents are missing and the next procedural step is unclear.

What I built: an intake and triage app that structures facts, identifies missing documents and prepares first-pass legal documents for human review.

Stack: Next.js, FastAPI, Firestore, Python, retrieval pipeline.

Status: showcase project.

• Partner, gunnercooke. Crypto, capital markets, payments and EU financial regulation.
• German-qualified lawyer and Volljurist, admitted to the German bar in 2012, with both state examinations.
• Associate-track experience at leading German and international law firms, including Hengeler Mueller, Freshfields Bruckhaus Deringer and Cleary Gottlieb Steen & Hamilton.
• Co-founder, AI-Lawyers, 2016.
• Data Scientist, justify.de and Dudenverlag. Built Python NLP pipelines for legal and lexicographic data.
• Conference background includes Blockchain Summit appearances in Amsterdam and Zurich.
• Languages: German and English.

• Firm: gunnercooke, Hamburg
• Firm contact: available via gunnercooke and LinkedIn
• LinkedIn: linkedin.com/in/sebastianforste
• GitHub: github.com/sebastianforste
• Email: sebastianforste@gmail.com