Boris Schickedanz

AI agents deliver more changes, but approvals remain a bottleneck. How companies can test new working rules and why starting afresh makes this easier.

Your Sunday read: Part 2 of "Why AI/SI needs second-order change" - can it be done in existing organizations?

Now, this is highly impressive.

Introducing AI/SI into an organisation can improve what it already does, but it can also slow down things. Realising its wider potential requires a willingness to change how the organisation itself works - the rules, processes and decision authorities.

The more I work with generative AI, the harder it gets to turn off my brain when watching Star Trek ... and wonder why the AI can do complex tasks, but not some of the most basic things. Of course, there will come a time when we still choose to do things manually that could easily be handled by AI  (cooking, painting, repairing things, maybe even coding). And for sure, it is fun to move up the sliders on the transporter console! 🥳 But when it comes to high-precision, fast-decision combat (sometimes the phasers are aimed by the AI, sometimes it’s just “enter firing pattern alpha”), navigating a nebula, or reconfiguring the warp core? Yes, we need the “human in the loop” in certain cases... but maybe not in life-threatening situations where milliseconds count. I know shows like Lower Decks do this for comedy, (cleaning the bio-filters works as a punishment, after all 🤮), and there is the “dramatic effect” needed for TV. But it really makes you think about where the line will be drawn between 'human growth', 'ethics' and 'AI efficiency'. And yes, probably AI will be banned from doing certain things, in combat as well. The question is: would other species do the same?

I am pretty sure AGI does not take a decade... with ideal circumstances, no politics, games or foul play involved I'd say within one year. AGI not in a sense that one AI can do what all humans combined can do (all at once), but with the capabilities of one average human. Lets give it 2-3 years from now.

So, is ChatGPT the new AppStore for AI? Well... they tried once, maybe as a test ballon (CustomGPT's)... I am pretty sure others are working on this too, especially @xai @elonmusk A bad idea would be to be locked in already... Maybe time to think of a cross-ai-platform language/platform/... (like Flutter,...)?

Nothing ever changes in business. Even when things are as clear as with generative AI and RAG...

Context Architecture: The Foundation of Enterprise AI Systems Why the future of AI implementation depends on treating context as infrastructure, not an afterthought When business leaders first encounter generative AI, they often think success comes from finding the right model or the perfect prompt. After implementing quite a few generative AI systems for enterprises, I’ve learned something different: the architecture of context is what separates AI experiments from AI systems that deliver lasting business value. Beyond Prompt Engineering: Building AI Context Architecture You’ve likely heard of “prompt engineering” – the practice of crafting better instructions for AI models. But what I am seeing in successful AI deployments goes far beyond writing clever prompts. We call it context architecture, and it’s fundamentally about architecture. Think of it this way: when you build a software system, you don’t just write code – you design databases, plan integrations, structure workflows, and create monitoring systems. Context architecture applies this same systematic thinking to AI systems. It’s the discipline of designing and optimizing all the information, instructions, and processes that surround your AI model to ensure reliable, measurable results and new value. The Hidden Infrastructure of AI Success Every successful AI system relies on carefully engineered context that most users never see. This invisible infrastructure includes: Structured Instructions: Not just telling the AI what to do, but precisely defining how it should approach problems, what format outputs should take, and how to handle edge cases. Dynamic Information Management: Automatically providing relevant, up-to-date context like current dates, company-specific data, or user preferences without overwhelming the system with unnecessary information. (think of RAG, memory, state and process descriptions for example) Quality Control Systems: Built-in processes to validate outputs, catch errors, and maintain consistency across different scenarios. Integration Architecture: Seamlessly connecting AI capabilities with your existing business systems, databases, and workflows. Why This Matters for Business Leaders Poor context architecture is often the root cause when AI initiatives fail to scale. I see companies that achieve impressive demos but struggle to deploy AI in production because they treated context as an afterthought rather than as foundational architecture. Consider a customer service AI system. A basic approach might simply tell the AI to “help customers with their questions.” But production-ready context architecture would include: Structured access to customer history and account information Instructions and descriptions of internal processes Clear escalation protocols for complex issues Brand voice guidelines and approved response templates Integration with ticketing systems and knowledge bases Continuous monitoring and improvement processes Our Approach to AI Architecture At Alpine Intelligence, our background in enterprise software architecture informs how we approach context architecture. Just as you wouldn’t build a critical business system without proper architecture, we believe AI systems require the same rigorous approach to context design. This means: Planning Before Building: Understanding your business requirements, data landscape, and integration needs before designing the AI system’s context architecture. Systematic Implementation: Creating structured, maintainable architecture that can evolve with your business needs rather than requiring constant manual adjustment. Measurable Outcomes: Building evaluation and monitoring into the context architecture from day one, so you can prove ROI and continuously improve performance. Knowledge Transfer: Ensuring your team understands not just how to use the AI system, but how its context architecture works and how to maintain it. Making Context Architecture Strategic The companies that will succeed with AI are those that treat context architecture as a strategic capability, not a tactical task. This requires: Investment in Architecture: Allocating time and resources to properly design context systems rather than rushing to deployment. Cross-Functional Collaboration: Bringing together technical teams, domain experts, and business stakeholders to inform context design. Continuous Optimization: Treating context as living infrastructure that requires ongoing measurement, refinement, and evolution. Internal Capability Building: Developing your team’s understanding of context architecture principles so they can maintain and improve AI systems over time. The Path Forward Context architecture represents the maturation of AI implementation from experimental to enterprise-grade. It’s the difference between AI systems that work in demos and AI systems that deliver sustained business value. For business leaders evaluating AI initiatives, the key question isn’t just “What can this AI model do?” but “How will we architect the context that makes this AI model consistently valuable to our business?” The companies that master context architecture will have a significant competitive advantage. They’ll deploy AI faster, more reliably, and with better business outcomes. More importantly, they’ll build AI capabilities that scale and evolve with their business rather than requiring constant manual intervention. Ready to explore how context engineering can accelerate your AI initiatives? Alpine Intelligence specializes in bridging the gap between AI potential and practical implementation through rigorous architectural approaches. Contact us to discuss your AI strategy and learn how proper context engineering can ensure your success

Yes, I can totally help with business cases enhanced or enabled with GenAI that work for you. 💡 Find out which ones will work & add value 🧑‍💻 Implementing them (or helping you implement them) 🧑‍🏫 Coaching your employees & devs on how to use the tools... 📧 Message me!

Die KI-Revolution als Katalysator menschlicher Evolution? KI entwickelt sich exponentiell weiter, das erleben wir alle täglich. LLMs schreiben inzwischen Code, Roboter übernehmen bald unsere Haushalte und intelligente Systeme treffen komplexe Entscheidungen. Doch während wir darüber diskutieren, dass enorme Effizienzgewinne und Automatisierung den Menschen überflüssig machen, übersehen wir eine wichtige Dimension dieser Entwicklung: KI ist nicht das Ende menschlicher Relevanz, sondern der Beginn einer notwendigen Evolution unserer Kapazitäten. Jenseits der Effizienzlogik: was wirklich zählt Die meisten Diskussionen über KI drehen sich um Produktivitätssteigerungen, Kosteneinsparungen und damit einhergehende Jobverluste. Das hat psychologisch betrachtet eine interessante Perspektive: KI fungiert als Spiegel, der uns zeigt, welche unserer Fähigkeiten tatsächlich mechanischer Natur sind und welche eine tiefere, nicht reproduzierbare Dimension besitzen. Algorithmen können Emotionen und Empathie simulieren – und das ziemlich überzeugend. Doch jeder, der gelernt hat, in sich hineinzuspüren, erkennt sofort den Unterschied zu authentischem menschlichem Mitgefühl. Diese Unterscheidungsfähigkeit haben wir alle, auch wenn sie bei vielen noch nicht bewusst entwickelt ist. Wenn rationale Intelligenz nicht mehr ausreicht... Unsere Gesellschaft hat jahrhundertelang rationale Intelligenz als höchste Form menschlicher Kapazität betrachtet. KI zeigt uns nun unwiderlegbar: Diese Form der Intelligenz ist reproduzierbar und sogar übertreffbar. Das mag erst mal ernüchternd sein, eröffnet aber die Möglichkeit, bisher vernachlässigte Dimensionen menschlicher Fähigkeiten zu erkunden. In der Geschäftswelt sehen wir bereits heute Führungskräfte, die über reine analytische Fähigkeiten hinausgehen und außergewöhnliche Ergebnisse erzielen. Sie treffen Entscheidungen, die sich nicht vollständig durch Datenanalyse erklären lassen, haben ein intuitives Verständnis für komplexe Systemdynamiken und inspirieren andere durch eine Präsenz, die sich jeder algorithmischen Reproduktion entzieht. Inspiration und Intuition als evolutionäre Intelligenz Echte Innovation entsteht selten durch reine Datenanalyse – das wissen wir eigentlich alle aus Erfahrung. Die bahnbrechendsten Durchbrüche entspringen einer Form der Intelligenz, die wir als Inspiration und Intuition bezeichnen können. Diese Kapazität ermöglicht es, Lösungen zu erkennen, die nicht durch logische Ableitung aus bekannten Parametern entstehen. Diese Intelligenz zeigt sich u.a. in der Fähigkeit, komplexe Herausforderungen ganzheitlich zu erfassen und Lösungsansätze zu entwickeln, die sowohl innovativ als auch (idealerweise) dem Gemeinwohl dienlich sind. Sie geht über die Begrenzungen des rein analytischen Denkens hinaus und eröffnet Zugänge zu kreativen Potenzialen, die in jedem Menschen angelegt sind. Viele erfolgreiche Unternehmer und Visionäre sprechen von Intuition, von "Bauchgefühl", von Momenten der Klarheit, die nicht rational und linear erklärbar sind. Wie entwickeln wir diese Fähigkeiten praktisch? Die Kultivierung dieser erweiterten Fähigkeiten erfordert eine bewusste Abkehr von der ausschließlichen Fokussierung auf rationale Optimierung. Es geht um die Entwicklung von Aufmerksamkeit für subtile Informationsebenen, die Schulung der Intuition und die Fähigkeit, aus einem Zustand innerer Ruhe heraus zu handeln. Konkret bedeutet das: Praktiken integrieren, die das Bewusstsein für non-lineare Denkprozesse schärfen, die Fähigkeit zur ganzheitlichen Problemwahrnehmung entwickeln und eine Offenheit für Eingebungen kultivieren, die Lösungen zum Wohle aller Beteiligten priorisiert (das ist wichtig, da dies scheinbar die Qualität der erhaltenen Informationen erhöht). Unternehmen, die das verstehen, schaffen Räume für kreative Reflexion, fördern werteorientierte Entscheidungsprozesse und investieren bewusst in die Entwicklung der emotionalen und intuitiven Intelligenz ihrer Mitarbeiter. Das ist übrigens nicht nur "nice to have" – das wird zum entscheidenden Wettbewerbsvorteil. Der evolutionäre Imperativ... warum wir keine Wahl haben Die KI-Revolution stellt uns vor eine grundlegende Wahl: Wir können versuchen, mit Maschinen in deren eigenem Spiel zu konkurrieren (spoiler alert: das werden wir verlieren), oder wir können diese Herausforderung als Katalysator für die Entwicklung authentisch menschlicher Potenziale nutzen. Die Zukunft gehört jenen Individuen und Organisationen, die KI als Werkzeug zur Befreiung menschlicher Kapazitäten verstehen. Während Algorithmen repetitive und analytische Aufgaben übernehmen, schaffen sie Raum für die Entfaltung kreativer, intuitiver und inspirativer Fähigkeiten. Eine erweiterte Definition von Intelligenz In einer Welt, in der rationale Intelligenz durch Maschinen übertroffen wird, müssen wir unsere Definition von Intelligenz grundlegend erweitern. Die Fähigkeit, innovative Lösungen zu inspirieren, komplexe ethische Entscheidungen zu treffen und andere zu transformierenden Handlungen zu motivieren – das sind Formen der Intelligenz, die sich der mechanischen Reproduktion entziehen. Diese erweiterte Intelligenz zeigt sich in der Kapazität, Visionen zu entwickeln, die sowohl realistisch als auch evolutionär sind, authentische Verbindungen zwischen Menschen zu schaffen und Lösungen zu erkennen, die über die Grenzen konventionellen Denkens hinausreichen. Letztendlich markiert die KI-Revolution nicht das Ende menschlicher Relevanz, sondern den Beginn einer bewussten Evolution unserer «höheren» Kapazitäten. Sie fordert uns auf, die Begrenzungen mechanischen Denkens zu überwinden und die inspirativen Potenziale zu entwickeln, die unser wahres Erbe als Menschen darstellen. Es wird Zeit, dass wir wieder zu dem werden, was wir eigentlich schon sind, nur scheinbar oft vergessen haben.

Case Study: The Intelligent Archive Using RAG and agentic workflow to access a knowledge base and auto-generate new solutions Initial Situation A company in the project development sector faced the challenge of optimally utilizing its extensive project documentation, market research data, and important correspondence. This valuable information was stored in a digital archive, but accessing and utilizing it was time-consuming. Alpine Intelligence was commissioned to develop an innovative solution that not only facilitates information access but also enables proactive support in core processes. 😰 Challenge The central challenges were: 🤯 Information Volume and Structure: The large volume of documents, often in unstructured formats like PDFs, made rapid and precise information retrieval difficult. 🤯 Contextual Understanding in Searches: Conventional search methods often did not provide sufficiently context-related results. 🤯 Utilization of Historical Data for New Projects: The potential to systematically learn from completed projects for new endeavors and to accelerate processes was not yet fully exploited. ⚡️ Solution by Alpine Intelligence⚡️ In an iterative process with the client, Alpine Intelligence designed and implemented a tailor-made solution based on Retrieval-Augmented Generation (RAG). Within the scope of our comprehensive Innovation Consulting, it was ensured that the solution was optimally tailored to the needs of the project developer. 💡The core components of the solution are: 🏛️ Intelligent Collection and Preparation of Documents in the Archive: Relevant texts are extracted from the data and converted into so-called vector embeddings using advanced AI models. These enable semantic, i.e., theme-based, similarity searches that go far beyond simple keyword searches. An automated process continuously collects new or changed documents in the archive. The documents are further enriched with additional data through analysis by an LLM, which is relevant for later queries but not semantically contained in the original document – this allows for a similarity search with a deeper understanding. 🏛️ Context-Related Optimization of Search Queries: The system understands and refines user queries by considering the conversation history and industry-specific terminology. This leads to significantly more relevant search results. 🏛️ From RAG to Agentic *: The system not only finds the most suitable documents but can also summarize the information contained therein into precise answers, cite the sources, and also generate more complex reports. RAG is just the beginning. Through an agent-based workflow, the system can also independently develop proposals or generate drafts for offers by accessing the collected knowledge in the archive and intelligently combining it – just as an employee would. Only the system is much faster. 🏁 Results The solution implemented by Alpine Intelligence led to significant improvements: 🚀Reduced Time Spent on Information Searches: Employees find needed information much faster and can concentrate on their core tasks. Onboarding new employees is also significantly easier. 🚀Improved Quality and Relevance of Information: The answers and search results are more precise and better tailored to the respective query. 🚀Increased Efficiency through Automation*: The ability to have drafts for offers or solution proposals created automatically significantly speeds up the corresponding processes. 🚀Higher User Satisfaction and Better Knowledge Management: Trust in the available information and the ability to draw from the entirety of the company’s knowledge have increased. 🚀Proactive Matching of Archive Knowledge with New Project Requirements*: The system can automatically provide relevant information from completed projects as soon as new project requirements are defined. 💡 Conclusion Through innovation consulting and the implementation of an advanced RAG solution with agent-based workflows, Alpine Intelligence has helped the project development company to revolutionize its internal knowledge management. This not only leads to a significant increase in efficiency and improved decision-making but also makes it possible to proactively use the collected know-how for shaping future success. * Agentic is currently in development