01
AI strategy & roadmap
A durable roadmap that does not fall apart in the first architecture review. Prioritised initiatives with owner, budget and quarter - built with someone who has shipped the systems underneath.
→
AI in production Access is governed.
Access is governed.
Sharing isn't.
An AI agent shares what it considers helpful - not what is permitted in context. Right person, wrong room. That is exactly what we control: Dissemination Control.
Based in Sweden - engagements across DACH, remote, on site for key meetings.
Industries from over two decades of engineering
- Rail transport
- Premium automotive
- Listed industrials
- Private banking
- Universal banking
- Asset management
- Medical technology
- Telecommunications
- Public safety
Practice
Five levers I usually pull.
I only take on work where I deliver more value than I charge. Engineering depth, not slideware.
02
Use-case discovery & prioritisation
A structured inventory across your functions. I score every use case on value, feasibility and risk - and tell you honestly which ones survive contact with production and which only look good in a workshop.
→
03
AI governance & EU AI Act
Policies, processes and roles that hold up under audit from August 2026 onward. Sized for mid-market reality, not DAX compliance overhead - including the case where data must never leave your infrastructure.
→
04
Training & enablement
From the leadership team to the developers who use AI every day. Formats that stick: ticket-based AI development, agent workflows, prompt and review discipline - not a one-off workshop forgotten after two weeks.
→
05
Process automation & agents
End-to-end automation where the maths works. I design it, build it with your team and operate it long enough to prove it - with governance and audit trails from day one. No lock-in.
→
Approach
Three phases. Nothing in between.
I
2–4 weeks
Diagnose
Interviews, code and architecture review, use-case inventory. Output is a shared picture of where AI moves the needle for you - and where it does not.
II
4–6 weeks
Design
Concrete initiatives, architecture options, a make-or-buy decision per case. We decide together; a steering committee signs off.
III
from 12 weeks
Deliver
I build and deliver the first one or two initiatives with your team - as lead architect or hands-on. Handover to a documented, operated, measured system.
The new reality
Capability used to be a question of company size. Not anymore.
With AI, a 100-person company now does things that would have taken a whole department five years ago: making sense of data, shipping faster, deciding better. What matters isn't the AI itself - it's whether it runs reliably, day in, day out, in real operation. That's exactly what I build. The projects below show how that looks.
Track record
Three projects. Decades of engineering behind them.
Selected engagements as lead architect, lead developer or senior business analyst - many of them turnaround mandates or critical first-time deliveries. I name them in conversation, under NDA.
Rail transport · 2021–2025
National rail operator, Germany
From monolith to microservices for a national rail operator
Card payment, ticket printing and signature capture, migrated from a monolithic application to 15+ microservices on Kubernetes/AWS. Over 100 REST APIs, 100+ ticket formats with precise layout control, millions of transactions in live operation. I led it as architect and lead developer - through every phase, from architecture through implementation to operation under high security requirements.
Outcome A monolith that slowed every change became a system carrying millions of transactions in live operation - maintainable, independently deployable, stable for years. What set the pace here wasn't the technology but the regulatory environment of a rail operator - exactly the brake that tends to be far smaller in a mid-sized company.
Spring Boot · Kubernetes/AWS · OAuth2/JWT · PostgreSQL · GitLab CI · Angular
Listed industrials · 2023–2024
Listed DAX industrial group
Workforce-planning platform with early LLM integration
Advisory and engineering for a workforce-planning programme with visibility up to top management. Data-processing logic for precise staffing decisions, a usable interface for genuinely complex HR data, a viable product strategy - and an LLM API integration where it actually paid off, not where it looked good on a slide.
Outcome Staffing decisions once buried in scattered spreadsheets now run through an interface that makes even the most complex HR data workable - with LLM support exactly where it pays off. A handful of people now make decisions at a quality that used to tie up an entire department.
Java · Spring Boot · Angular · Kubernetes · MongoDB · LLM-API
Premium automotive · 2018–2019
European premium car manufacturer
Music-streaming integration into a connected-car system
Concept and technical delivery of a global music-streaming provider's integration with the vehicle infrastructure. Direct alignment with the streaming partner, leadership of an international team, and accountability for timeline and budget.
Outcome Concept became a shipped feature in the vehicle - a clean interface to a global tech partner, aligned across borders, on time and on budget. Proof that integrating with a corporate giant needn't be a big-project risk when it's scoped right.
Apple Music API · RESTful Microservices · Java · Spring Boot · AngularJS · Git · Bitbucket · Google Cloud Platform (GCP) · AWS · Microsoft Azure · Jira · Confluence · Scrum · Kanban · Agile methods · DevOps
About
Andre Jahn
Independent AI Advisor · Jahn Consulting
An IT consultant, software architect and lead developer for German corporates and the mid-market for over two decades. No formal AI qualification - a career changer who learned AI by building production systems with it, not in a lecture hall.
My specialty is the projects on the brink of failure: turnarounds, migrations two other teams couldn't finish, systems that simply have to go into production. Before I came to AI, I built mission-critical software for rail payment, FDA-regulated medical data, public safety, premium automotive and private banks.
Today I bring that engineering discipline into AI projects. Taking GenAI from pilot to production is a technical discipline, not a rhetorical one - and that is exactly where most projects get stuck.
" No MBA, no slideware. Engineering depth is my only lever - and that is exactly the point.
Insights
What I'm currently publishing.
Notes from practice - and what I open-source.
Latest post · AI Agents
Recent skill.md Is a Supply Chain
Stack Overflow for Agents ships one of the most careful skill.md files out there - and that's exactly why it exposes a structural problem in the whole skill economy: mutable remote instruction sources without integrity. A close read of 692 lines and why it takes technical controls, not prose.
Read the post →
Open-source blueprint Dissemination Control for AI Agents
AI agents share what they consider helpful - not what is permitted in context. Dissemination Control closes that gap: vendor-neutral, building on your existing IAM, validated in a live reference environment. On GitHub, CC-licensed.
Read the blueprint on GitHub ↗
-
What Remains When the Model Is Gone
22 Jun 2026 -
What Does Your AI Know About Your Business? Hopefully Nothing.
15 Jun 2026 -
Free-Text Fields from Hell
14 Jun 2026
Contact Let's talk about what
Let's talk about what
you're actually trying to do.
A 45-minute introduction - no sales pressure, no slides. I listen, give a first read, and tell you openly whether I'm the right partner.
Book an introduction →- GitHub
- github.com/JahnConsulting