$1.200 Fullstack Course Free With a Membership!

Strategic AI Development: Beyond Hype to Scalable Impact

Leveraging the best sessions from MLcon to power your 2026 roadmap.

Context instead of hype – truly understanding AI

Artificial intelligence is evolving at an unprecedented pace. New models, frameworks, and buzzwords appear almost daily. Yet while technologies change, the core principles of successful AI systems remain the same: sound architecture, deep context understanding, and thoughtful strategic decisions.

This course distills the most valuable insights from leading MLcon sessions into a long-lasting knowledge foundation for developers and AI professionals. Instead of a year-specific recap, you gain a clear understanding of the concepts, patterns, and mental models that will stay relevant in 2026 and beyond.

From Experimentation to Production-Ready AI

Modern AI is about more than experimenting with the latest models. Real impact emerges when LLMs, agents, data, architecture, and governance work together as a cohesive system. In this course, you’ll learn:

why many AI initiatives fail despite powerful models
how context, evaluation, and system design shape AI quality
the role of embeddings, RAG, agents, and protocols in scalable solutions
how to design AI systems that are robust, explainable, and maintainable

The content is intentionally model-agnostic, helping you make better decisions regardless of whether you work with GPT-4, GPT-5, or open-source alternatives.

Structured outputs in Open AI | Rainer Stropek, Melanie Bauer

OpenAI’s language models excel at generating fluent text, but their impact is magnified when they produce outputs in a predetermined JSON structure. In this session, Rainer Stropek will explore the capabilities of Structured Outputs-techniques that ensure AI responses adhere to developer-defined schemas. Attendees will gain insights into how structured output enabling robust, production-ready integrations of OpenAI into custom application. Practical examples will illustrate how structured outputs can enhance reliability and reduce post-processing efforts. A basic understanding of OpenAI APIs is recommended to fully appreciate the applications discussed in this talk.

How to Make an LLM Truly Understand Your Context | Peter Fuchs

Discover how to achieve a significant improvement in answer quality in Retrieval Augmented Generation (RAG) systems. Common RAG solutions are known to combine the strengths of retrieval-based methods and generative models to use the power of generative AI with company data. However, many of them fail to deliver the required accuracy and relevance of generated text as they consist of standard components and architecture. By leveraging advanced techniques like knowledge graphs, you can greatly surpass basic RAG implementations. This approach not only enhances performance but also makes the specific quality of the LLM less critical and reduces hardware requirements, optimizing both results and resource utilization.

AI Agents Unplugged: No Frameworks, Just Code | Paul Dubs

This talk provides a practical, step-by-step guide to building AI agents using only Python and an LLM API. Starting from a basic API call, we’ll progressively add capabilities: prompt engineering for consistent outputs, implementing memory and context management, adding tool use and function calling, handling errors and retries, and implementing basic reasoning loops. Along the way, we’ll identify common pitfalls including context window limitations, hallucination in tool use, infinite loops, and error cascades. Each implementation choice will be explained with working code examples. By the end, attendees will understand the core components of AI agents and have the knowledge to build their own agents for real-world tasks without relying on frameworks or abstractions.

Unlocking Strategic Insight with Unified Data Repositories and AI Copilots | Sinda Khenine

Organizations today struggle with fragmented data silos across departments and global sites. We face this daily – working with scattered datasets ranging from structured test results to unstructured consumer feedback and financial reports. This fragmentation leads to duplicated efforts, disconnected insights, and missed opportunities. Our approach consolidates diverse datasets into a unified, secure, globally accessible repository. On top of this, we integrate an AI copilot to answer complex, cross-domain questions and uncover hidden patterns. However, not all problems require full AI. We selectively use machine learning for tasks like anomaly detection or forecasting, while relying on simpler analytics where they are more cost-effective. Key challenges include securing sensitive data, preprocessing for integration, ensuring global accessibility, and balancing performance with cost. Building such a system means making practical trade-offs – deciding where AI adds value and where lightweight ML or rule-based tools suffice. This session shares strategies, lessons, and implementation insights from our journey. Attendees will learn how to transform fragmented data into strategic assets, using the right mix of technology to enhance decision-making while maintaining scalability, security, and efficiency.Join us to explore how intelligent data consolidation can drive meaningful innovation and business value.

MCP, A2A, and AG-UI: Architecting Agent Intelligence Beyond Chat | Max Marschall

Agentic systems are emerging as the next paradigm in AI application design – moving beyond static chatbots toward dynamic, context-aware, and modular ecosystems of intelligent agents. To operate effectively at scale, these systems must integrate three foundational capabilities: perceiving their environment, collaborating across specialized components, and interacting with users in real time. This talk introduces an architecture that leverages three interoperable protocols to achieve these goals. The Model Context Protocol (MCP) enables dynamic context hydration and semantic grounding, allowing agents to operate on structured and unstructured inputs tailored to specific tasks. The Agent-to-Agent Protocol (A2A) facilitates orchestrated collaboration between modular agents, enabling delegation, specialization, and distributed reasoning. The Agent-User Interaction Protocol (AG-UI) provides a real-time interface layer that closes the loop with users, supporting direct feedback, steerability, and reactive experiences. Together, these protocols form a cohesive foundation for building intelligent systems that are decoupled, composable, and resilient – minimizing integration debt while enabling rich, responsive behaviors. This session explores design patterns, practical challenges, and architectural strategies for applying MCP, A2A, and AG-UI in real-world agentic applications, offering a blueprint for anyone aiming to architect intelligence beyond chat.

Breaking down an LLM for its parts | John Davies

AI has been around for many years, we’ve been using text to speech and speech to text for over a decade, Siri, Alexa etc., cameras could follow faces a decade ago. However, the major change was just two years ago when Generative Pre-trained Transformers emerged for public use, yes GPT. These are Large Language Models (LLMs) but just for us geeks, are the open source LLMs like Phi 3.5, Qwen 2.5, Llama 3.2 and Mistral, these are advancing so fast I fear even this abstract will be out of date. John will pull these LLMs apart to show the internal workings, demonstrate some cool features and help you better understand how they work, what they’re good at and what they’re not good at and why. From vocabulary, tokenisation and embeddings to attention heads, quantisation and performance. We’ll be running everything locally and will try some German in the LLMs too, simple code yet fascinating results. If you get change to download “ollama” (.com) and one or more of the models mentioned above on your laptop, please bring it along.

Beyond LLMs: Using Embedding Models for Input Guarding, Semantic Routing, and Tool Decisions | Marco Frodl

Retrieval Augmented Generation (RAG) leverages retrievers like vector databases to fetch relevant data for answering queries. In advanced RAG setups involving multiple data sources, selecting the best retriever is critical. Traditionally, in LangChain this is handled by a MultiRoute Chain, where a Large Language Model (LLM) dynamically chooses the optimal data source based on semantic fit. However, this approach can be slow, costly, and unpredictable. Enter Semantic Router—a faster, cheaper, and deterministic alternative that uses an embedding model for retriever selection without compromising quality. In this talk, I’ll showcase the Semantic Router’s broader capabilities, including input guarding for AI applications and efficient tool selection for function calling. Through live coding, we’ll first build a traditional MultiRoute Chain and then optimize it with Semantic Router, illustrating how this transformation can dramatically improve efficiency in RAG workflows.

Demystifying Deep Learning | Rachel-Lee Nabors

What do acronyms like RAG and GPT really mean? What is the math behind the “magic” of deep learning? What is a “perceptron” and why is it called a “neural network?” This talk demystifies AI’s field of machine learning for AI-curious technical and non-technical audiences alike. The concepts, mental models, and history needed to grok are colorfully illustrated and explained. From the award-winning cartoonist turned engineer/keynote speaker who has taught millions of developers via react.dev and MDN.

Take Your Skills to The Next Level

Join MLcon London

Create & innovate with Generative AI, LLMs & Machine Learning, at MLcon London. Turn theory into action and learn to build AI-powered intelligent systems from industry experts. Deep dive into Advanced ML & MLOps—from prototype to production. Book your tickets and join us next May 11 – 15, 2026!

RESERVE YOUR SPOT!

Expert knowledge for…

AI Engineers & ML Practitioners who want to make their systems more robust, traceable and production-ready
Software developers who want to meaningfully integrate AI into existing architectures
Tech professionals and architects who need to understand AI not only from a technical but also from a strategic perspective.

Join us and learn...

To thoroughly evaluate and classify current AI and ML approaches
Develop more intelligent systems with a clear architecture and clean context structure
Better justify technical decisions regarding LLMs, evaluation and integration
Plan AI projects strategically instead of just implementing them experimentally.
Transfer insights from MLcon 2025 directly into your own work