# Alexander C.S. Hendorf

> Independent AI & Open Source Strategist — Solo Consultant

I bring clarity before AI initiatives lock in vendors, platforms, and investments. I work personally with your decision-makers and engineering team. No team, no account manager — full accountability from one strategist.

## Services

- [AI Strategy with Execution Foundation](https://hendorf.com/en/#services): For decision-makers who need robust clarity before committing to an investment.
- [Execution Governance for AI Transformation](https://hendorf.com/en/#services): For leaders who need to de-risk execution. I steer, your team delivers.
- [Interim Leadership Mandate](https://hendorf.com/en/interim-ai-leadership/): For organisations that need operational AI or data leadership on an interim basis.

## Current Research: The Governance Gap

Companies believe they are adopting AI. In fact, they are quietly shifting responsibility onto their developers. I turn that shift into a decision, so you can scale AI without leaving responsibility, ROI and risk ungoverned.

71% say accountability for AI-caused defects is unregulated and lands on the developer; 40% name unclear governance as the No. 1 bottleneck, ahead of technology; 6% feel relieved at day's end; 67% experience denser, more demanding days. Source: Python Software Association Germany · a survey of 383 software professionals · June 2026.

Framework: Accountability (Who signs off AI code, and does that hold up in an audit?); ROI reality (Does the speed still pay after review, rework and risk?); Value migration (Where does the value of your engineering move, and does the role model follow?); Deskilling (In three years, will your organisation still understand its own code?); Shadow AI (Who decides on tools that are already in use?).

Offering: [Boardroom-to-Code Session](https://hendorf.com/en/boardroom-to-code-s26/) — Diagnostic → Decision → Operating model.

## Case Studies

- **Quantitative Asset Management**: Legacy C# and SAS silos separated research, portfolio management, and engineering. Long release cycles, limited ESG capability, and a lack of traceability put speed, control, and compliance at risk. → Migration of the organisation to Python and open source. Deployment cycles dropped from three months to three weeks; 44 employees were upskilled across four cohorts. The result was a stack that supports traceability, auditability, and regulatory requirements in a financial-services environment.
- **Public Infrastructure**: More than 30 years of project data, 70% manual processing, and no reliable data foundation. Publicly funded information was trapped in silos; AI potential could not be put to operational use. → Development of a data strategy with a 120-page implementation roadmap. An AI-supported forecasting model for workforce planning achieved 90% accuracy; the planning cycle shifted from annual to continuous.
- **Automotive / Research and Development**: More than 10,000 research documents spanning three decades: unstructured, confidential, and nearly impossible to search. Cloud and API usage were ruled out. → Built an NLP-based knowledge explorer that turned static document storage into search results in seconds. Topics, related contexts, and source documents became directly accessible — fully on-premises, with no external APIs and no LLMs. The internal team subsequently expanded the solution further.

## Glossary

Curated AI glossary for decision-makers in regulated industries. Each entry: definition, vendor-narrative vs. technical/regulatory reality, and the right boardroom question. (31 entries)

- [Agentic AI](https://hendorf.com/en/glossary/agentic-ai/): AI systems that autonomously plan and execute multi-step tasks across multiple tool calls.
- [AI Act (EU AI Act)](https://hendorf.com/en/glossary/ai-act/): EU Regulation 2024/1689 governing AI systems, with staggered application through 2027 and separate obligations for general-purpose AI models.
- [AI Red Teaming](https://hendorf.com/en/glossary/ai-red-teaming/): Structured, adversarial testing of an AI system against security, bias, hallucination and misuse patterns.
- [Context Window](https://hendorf.com/en/glossary/context-window/): Maximum number of tokens a language model can process simultaneously per request.
- [Differential Privacy](https://hendorf.com/en/glossary/differential-privacy/): A mathematical privacy standard that bounds the influence of any single record on a computation by a quantified parameter (epsilon).
- [Distillation](https://hendorf.com/en/glossary/distillation/): A technique for training a smaller model to approximate the behaviour of a larger one — at significantly lower inference cost.
- [DORA (Digital Operational Resilience Act)](https://hendorf.com/en/glossary/dora/): EU Regulation 2022/2554 on the digital operational resilience of financial entities, applicable since 17 January 2025.
- [Edge AI](https://hendorf.com/en/glossary/edge-ai/): Running AI models directly on end devices instead of in central data centres or cloud APIs.
- [Embedding](https://hendorf.com/en/glossary/embedding/): A numeric vector representation of a text, image or data object in a high-dimensional space, where semantically similar objects sit close together.
- [Evaluation (Eval)](https://hendorf.com/en/glossary/evaluation/): Systematic, reproducible measurement of an AI system's quality against defined criteria.
- [Federated Learning](https://hendorf.com/en/glossary/federated-learning/): A training approach in which a shared model is trained across multiple decentralised data sources without the training data leaving those sources.
- [Fine-Tuning](https://hendorf.com/en/glossary/fine-tuning/): A technique for retraining the weights of a pre-trained foundation model on a domain-specific dataset.
- [Foundation Model](https://hendorf.com/en/glossary/foundation-model/): A large, generally pre-trained AI model that serves as the basis for a variety of downstream applications.
- [Guardrails](https://hendorf.com/en/glossary/guardrails/): Mechanisms before, during or after model inference that filter, restrict or escalate undesired inputs or outputs.
- [Hallucination](https://hendorf.com/en/glossary/hallucination/): Output of a language model that is plausibly worded but factually wrong or not supported by the sources.
- [Harness](https://hendorf.com/en/glossary/harness/): A structured software layer around an AI model that orchestrates tool calls, eval routines, guardrails and output processing.
- [Inference Cost / TCO](https://hendorf.com/en/glossary/inference-cost-tco/): Ongoing cost of model use per request; TCO extends this to development, eval, hosting, monitoring and compliance over the lifecycle.
- [Loop Engineering](https://hendorf.com/en/glossary/loop-engineering/): Instead of prompting an AI agent step by step, you design the system that prompts it — one that finds, dispatches, checks and remembers work on its own.
- [Mixture of Experts (MoE)](https://hendorf.com/en/glossary/mixture-of-experts/): An architectural pattern in which a model consists of several specialised subnetworks, only a small selection of which is activated per token.
- [Model Card](https://hendorf.com/en/glossary/model-card/): A structured document describing the training data, intended use, limitations, bias patterns and licence of an AI model.
- [Model Governance](https://hendorf.com/en/glossary/model-governance/): Processes, roles and documentation that steer the lifecycle of an AI model in a company.
- [Multimodal](https://hendorf.com/en/glossary/multimodal/): An AI model's capability to process multiple input and output modalities — typically text, image, audio and video.
- [On-Premises AI](https://hendorf.com/en/glossary/on-premises-ai/): Operating AI models and infrastructure in your own or a dedicated environment instead of through the model providers' API services.
- [Open Weights vs. Open Source](https://hendorf.com/en/glossary/open-weights-vs-open-source/): Open weights denotes publication of the model parameters; open source additionally requires training code, data specification and an open licence.
- [Prompt Injection](https://hendorf.com/en/glossary/prompt-injection/): An attack technique in which inputs are crafted so that the model ignores or overrides its original system instructions.
- [RAG (Retrieval-Augmented Generation)](https://hendorf.com/en/glossary/rag/): An architectural pattern in which a language model retrieves relevant documents from a knowledge base and integrates them as context in the prompt.
- [Reasoning Model](https://hendorf.com/en/glossary/reasoning-model/): A class of language models that produce a longer internal chain of thought before answering and outperform classical LLMs on multi-step tasks.
- [Sovereign AI](https://hendorf.com/en/glossary/sovereign-ai/): AI infrastructure under national or European control — across data, operations, models and training data.
- [Synthetic Data](https://hendorf.com/en/glossary/synthetic-data/): Artificially generated data that reproduces statistical or structural properties of real data.
- [Tokenmaxxing](https://hendorf.com/en/glossary/tokenmaxxing/): Status game in which engineers and teams compete on raw token consumption — often treated as a proxy for productivity rather than a measure of outcome.
- [Vector Database](https://hendorf.com/en/glossary/vector-database/): A database optimised for storage and search of high-dimensional vectors (embeddings).

## Credentials

- German AI Association | Head of Open Source Working Group
- Python Software Foundation Fellow
- Pioneers Hub Initiator
- PySV Board Member
- 50+ technologies deployed across enterprise and open-source ecosystems
- 100+ conference talks in 15+ countries (ESA, UN AI for Good Geneva, WHU, PyData, EuroPython)
- Featured in Handelsblatt, ZEIT, Tagesspiegel
- Company: opotoc GmbH

## Speaker Topics

- Agentic AI: From hype to value
- AI & open source in regulated industries
- AI strategy beyond tools

## Blog

- [The New AI Architecture in 2026: Harness, Evaluation, Open Source](https://hendorf.com/en/blog/ai-architecture-2026-harness-evaluation-open-source/): Three architectural shifts separate 2026 programmes that ship from those on 2024 logic: model to harness, test to architecture, budget to sovereignty lever.
- [Mastery, not Ownership: Where AI Sovereignty Actually Takes Shape in 2026](https://hendorf.com/en/blog/open-source-sovereignty-2026/): Six axes where it shows what actually runs sovereign: data, inference, post-training, evaluation, compliance, operations.
- [Regulated, Interconnected, Stalled: What's Blocking AI Projects in Five Industries](https://hendorf.com/en/blog/problem-clinic-pyconde-2026/): Insights from a closed-door session under Chatham House Rule: the hardest blockers sit ahead of the technology—missing shared language, standard, mandate.
- [Evaluation beats architecture: the discipline that decides whether AI ships](https://hendorf.com/en/blog/evaluation-is-architecture/): The most important architectural decision for LLM systems in 2026 isn't the model: it's evaluation. Three shifts separating production AI from stalled pilots.
- [Agents in Production: Why Evaluation Matters More Than Model Choice](https://hendorf.com/en/blog/agents-in-production-2026/): At PyCon DE & PyData 2026 the agent talks turned sober: less model magic, more harness, context budgets, deterministic fallbacks, eval-coupled releases.
- [Stop Waiting, Start Shipping: the Open AI Stack Grew Up in 2026](https://hendorf.com/en/blog/stop-waiting-start-shipping/): Sebastian Raschka in a fireside chat: why the harness matters more than the base model, and why experienced engineers become more valuable in 2026.
- [Agentic AI & Automation: Three Months of Coding Agents in Operations](https://hendorf.com/en/blog/agentic-ai/): Three months of coding agents in PyCon DE & PyData ops: where they hold up, where they fail, and why governance matters more than tool choice.
- [Cross-Pollination & AI: Scaling in the Era of Autonomy](https://hendorf.com/en/blog/cross-pollination-ai/): AI is permeating every layer of the organisation. The companies that win in the era of autonomy are the ones with the best cross-pollination across boundaries.
- [Enterprise AI & Open Source: What Three Practitioners Said About a Sustainable Architecture](https://hendorf.com/en/blog/enterprise-ai-open-source/): From a fireside chat with Merck, Quoniam and Explosion/spaCy: open source as the pragmatic enterprise default, standardisation as the high-leverage investment.
- [Data Quality Assurance: The Foundation for Reliable AI Systems](https://hendorf.com/en/blog/data-quality-assurance/): Data quality decides whether AI systems hold up in production. Patterns from enterprise mandates, briefed for science-spinout founders.
- [Overfitted Promises: AI in Coding Research – Hype vs. Evolution](https://hendorf.com/en/blog/overfitted-promises/): Today's AI coding tools work more like alchemy than established science: that's not a verdict, it's a stage. Where they pay off, where they fail.
- [Why AI Projects Fail](https://hendorf.com/en/blog/why-ai-projects-fail/): AI projects rarely fail on the technology. They fail on people, organisations and missing governance: three acts and the levers that work instead.
- [The Economics of Prediction Machines: Understanding AI as an Economic Factor](https://hendorf.com/en/blog/economics-of-prediction-machines/): AI is not IT. It is R&D. The companies that win as prediction costs collapse treat prediction capacity as strategy, not as a technology purchase.
- [AI for Decision-Makers: Why You Can't Buy AI, You Have to Build It](https://hendorf.com/en/blog/ai-for-decision-makers/): AI is not a product you buy: it's an organisational and technological capability you build. Six years of LLM acceleration haven't changed that.
- [What Two Years of Deep-Learning Experiments Teach About AI in Practice](https://hendorf.com/en/blog/deep-learning/): Two years of hands-on PyTorch experiments, and the unglamorous lessons about data quality, the 90% trap and the real gap between lab and production.
- [Agile Analytics Rock the Enterprise: Open Source as a Game-Changer](https://hendorf.com/en/blog/agile-analytics-enterprise/): A 2018 insurance case study: out of proprietary silos into a Python-based open-source stack, with change management the tech alone cannot provide.

## Links

- Website: https://hendorf.com
- LinkedIn: https://linkedin.com/in/hendorf
- GitHub: https://github.com/alanderex
- Speaker page: https://hendorf.com/en/speaker/
- Contact: https://hendorf.com/en/#contact