TL;DR: In March I argued the CLI vs MCP debate was the wrong debate, and that the CLIโs advantages were a temporary artifact of training data, not a law of physics. One of those advantages was multi-aโฆ
Architecture
44 posts tagged Architecture ยท all tags
2026
CLI vs MCP, Part Two: The First Gap Just Closed
Context Engineering: The Skill That Replaced Prompt Engineering
TL;DR: Patrick Debois coined DevOps in 2009 by naming what practitioners were already doing. In 2026, heโs doing it again with โContext Engineeringโ and the CDLC (Context Development Lifecycle): Generโฆ
Cognitive Debt: The Hidden Cost of AI-Generated Code
The Code Nobody Understands Hereโs a pattern Iโve seen across multiple teams: a data pipeline ships, built almost entirely by an AI coding agent. Clean architecture. Full test coverage. Passes every rโฆ
8 AWS Guides for Agentic AI โ Mapped to the 4 Pillars That Get You to Production
The Gap Between Demo and Deployment TL;DR: AWS released 8 prescriptive guides for building production-ready agentic AI. This post maps each guide to the four pillars that get agents from demo to deploโฆ
Intelligence Is About Time, Not Parameters
The Question Every SA Gets Beyond a complexity threshold, larger models become less insightful โ the savant regime. โWhich model should I use?โ I hear it in almost every customer conversation about geโฆ
What Reasoning Actually Means (and Why It Matters for Your Architecture)
It Started with a Saturday Morning Experiment I recently ran a simple test. I asked a small language model the same questions three times, with zero, one, and three rounds of self-reflection, and publโฆ
Is the AI Subsidy Era Ending? And Why That Might Be a Good Thing
I was listening to a recent episode of The AI Daily Brief โ โThe AI Subsidy Era Is Overโ โ and my thoughts started spinning. Not because the argument was new, but because it connected dots Iโd been seโฆ
The Agent Security Stack Nobody Is Building
The Scenario Nobody Planned For Itโs 11 PM. Your customer support agent, the AI one, is processing a refund request. It queries the order database, pulls the customerโs payment history, and calls the โฆ
Software Fundamentals Matter More Than Ever
The Talk That Confirmed What Iโve Been Seeing Matt Pocock stood on stage at the AI Engineer Summit and said something that most of the audience needed to hear: the developers who succeed with AI codinโฆ
MCP Sampling & Elicitation: When Servers Talk Back
From Request-Response to Collaboration When I wrote about the CLI vs MCP debate [1], I focused on the infrastructure patterns underneath. But MCP itself has been evolving, and the latest additions chaโฆ
Nvidia's Real Moat: What Jensen Huang Told Dwarkesh Patel
Electrons In, Tokens Out Long weekend drive, sunny weather, and nearly two hours of Jensen Huang arguing with Dwarkesh Patel about whether Nvidiaโs moat will hold. As far as podcast entertainment goesโฆ
Self-Improving Models: What MiniMax M2.7 Actually Does
The Headline vs The Reality โModel trains itself over 100+ autonomous cycles.โ That was the headline when MiniMax released M2.7 on March 18, 2026 [1]. It sounds like science fiction: a model bootstrapโฆ
The Citation Crisis: What AI Hallucinations Mean for Your Enterprise
The Reference I Almost Didnโt Check A few days ago, I was reviewing an article my AI agent had drafted. The sources section looked clean: numbered references, proper formatting, plausible titles. One โฆ
From Cloud-Native to AI-Native: What Actually Changes
The Fifteen-Year Echo Fifteen years apart. Same stage. Different world. In 2010, Adrian Cockcroft stood on the QCon stage and told the audience that Netflix was running its entire business on a publicโฆ
The Protocol We Should Have Built for Humans
Namaste from 6,165 Meters I just summited Imja Tse (Island Peak, 6,165 meters) in Nepal. No Slack, no email, no MCP servers crashing in the background. Just ice, thin air, and the kind of clarity thatโฆ
Is RAG Still Needed with 1M+ Token Context Windows?
The Kofferklausur, Revisited In September 2024, a colleague asked an audience: โWhat is RAG?โ I answered: Kofferklausur [1]. For non-German speakers: a Kofferklausur is an open-book exam. You bring yoโฆ
LLMs Don't Do Math โ They Predict What Math Looks Like
The Invisible Error To test this, I designed five calculations that anyone in business might ask an AI assistant, the kind of questions youโd type into ChatGPT or Claude expecting a quick, reliable anโฆ
Your AI Models Have an Expiry Date โ A Practical Guide to Model Lifecycle Management
Introduction โ The Promise I Made In my previous article [1], I explored the maintenance trap in IT โ how software systems are more like plants than stones, requiring constant care. I ended with a cliโฆ
Most comprehensive overview on RAG I have seen. We came a long way from vanilla RAG. Still remember
Most comprehensive overview on RAG I have seen. We came a long way from vanilla RAG. Still remember the time of arguments that RAG is just a โhot fixโ to be obsolete soon. Reality is it is not a fix bโฆ
IT System Maintenance in the age of AI
IT System Maintenance in the age of AI Introduction - The Maintenance Trap in IT You donโt need to be in the IT industry for long to have witnessed this firsthand. Even non-IT users do. Those systems โฆ
๐ง ๐ง๐ต๐ฒ ๐ ๐ฎ๐ถ๐ป๐๐ฒ๐ป๐ฎ๐ป๐ฐ๐ฒ ๐ง๐ฟ๐ฎ๐ฝ: ๐ช๐ต๐ ๐ฌ๐ผ๐๐ฟ ๐๐ง ๐ฆ๐๐๐๐ฒ๐บ๐ ๐๐ฟ๐ฒ ๐ ๐ผ๐ฟ๐ฒ ๐๐ถ๐ธ๐ฒ ๐ฃ๐น๐ฎ๐ป๐๐ ๐ง๐ต๐ฎ๐ป ๐ฆ๐๐ผ๐ป๐ฒ๐
๐ง ๐ง๐ต๐ฒ ๐ ๐ฎ๐ถ๐ป๐๐ฒ๐ป๐ฎ๐ป๐ฐ๐ฒ ๐ง๐ฟ๐ฎ๐ฝ: ๐ช๐ต๐ ๐ฌ๐ผ๐๐ฟ ๐๐ง ๐ฆ๐๐๐๐ฒ๐บ๐ ๐๐ฟ๐ฒ ๐ ๐ผ๐ฟ๐ฒ ๐๐ถ๐ธ๐ฒ ๐ฃ๐น๐ฎ๐ป๐๐ ๐ง๐ต๐ฎ๐ป ๐ฆ๐๐ผ๐ป๐ฒ๐ After years of watching organizations struggle with outdated systems, Iโve written about a pattern we all know too wellโtheโฆ
๐ฏ 'How do we pick the RIGHT AI agent use case?
๐ฏ โHow do we pick the RIGHT AI agent use case? This is the question I hear most from customers exploring agentic AI. Hereโs the mechanism I run through together with the customer: The 4-Quadrant Evaluโฆ
2025
๐ฃ๐ฒ๐ฟ๐ณ๐ฒ๐ฐ๐ ๐๐ถ๐บ๐ถ๐ป๐ด ๐ณ๐ผ๐ฟ ๐ฎ๐ฌ๐ฎ๐ฒ ๐ฐ๐ฎ๐ฟ๐ฒ๐ฒ๐ฟ ๐ฝ๐น๐ฎ๐ป๐ป๐ถ๐ป๐ด! ๐ฏ
๐ฃ๐ฒ๐ฟ๐ณ๐ฒ๐ฐ๐ ๐๐ถ๐บ๐ถ๐ป๐ด ๐ณ๐ผ๐ฟ ๐ฎ๐ฌ๐ฎ๐ฒ ๐ฐ๐ฎ๐ฟ๐ฒ๐ฒ๐ฟ ๐ฝ๐น๐ฎ๐ป๐ป๐ถ๐ป๐ด! ๐ฏ I just dived deep into the book โSolutions Architect Interview: Winning strategies and effective tactics for interview successโ by Saurabh Shrivastava, Neelaโฆ
๐ค ๐ช๐ต๐ฎ๐ ๐ถ๐ ๐๐ผ๐๐ฟ ๐๐ ๐ฆ๐๐ฟ๐ฎ๐๐ฒ๐ด๐? Chasing single point solutions or exploring system-level AI solutions? Yesterday I had the pleasure of listening to a great presentation by Chris Nosko. Among other importaโฆ
๐ฏ ๐๐ฒ๐ฒ๐ฝ ๐ฑ๐ถ๐๐ฒ ๐ถ๐ป๐๐ผ ๐๐ฒ๐น๐น-๐๐ฎ๐๐ฒ๐ฑ ๐๐ฟ๐ฐ๐ต๐ถ๐๐ฒ๐ฐ๐๐๐ฟ๐ฒ๐
๐ฏ ๐๐ฒ๐ฒ๐ฝ ๐ฑ๐ถ๐๐ฒ ๐ถ๐ป๐๐ผ ๐๐ฒ๐น๐น-๐๐ฎ๐๐ฒ๐ฑ ๐๐ฟ๐ฐ๐ต๐ถ๐๐ฒ๐ฐ๐๐๐ฟ๐ฒ๐ Last week I attended an outstanding presentation by my colleague Robert Himmelmann on โCell-based Architecturesโ โ one of the most insightful deep-dives Iโve โฆ
๐๐ฒ๐ป๐ฒ๐ฟ๐ฎ๐๐ถ๐๐ฒ ๐๐ โ ๐ง๐ต๐ฒ ๐ฃ๐ฒ๐ป๐ฑ๐๐น๐๐บ ๐๐ฒ๐ฒ๐ฝ๐ ๐ฆ๐๐ถ๐ป๐ด๐ถ๐ป๐ด?!
๐๐ฒ๐ป๐ฒ๐ฟ๐ฎ๐๐ถ๐๐ฒ ๐๐ โ ๐ง๐ต๐ฒ ๐ฃ๐ฒ๐ป๐ฑ๐๐น๐๐บ ๐๐ฒ๐ฒ๐ฝ๐ ๐ฆ๐๐ถ๐ป๐ด๐ถ๐ป๐ด?! Enterprises have long followed a familiar rhythm. A major consulting firm arrives, declares centralization the new path to efficiency; a few years later, โฆ
If you don't have the data available, implementing an AI use case becomes a data
If you donโt have the data available, implementing an AI use case becomes a data gathering death march, often crossing organizational boundaries. Instead of spending 80% of the project time on buildinโฆ
I just signed up for the Software Architecture Superstream: Architecture Patterns and Antipatterns f
I just signed up for the โSoftware Architecture Superstream: Architecture Patterns and Antipatterns for AIโ[1] which is taking place at 12th August CEST late afternoon. The lineup of speakers and topiโฆ
Diving into designing multi-agent systems and got lost with all the different implementation options
Diving into designing multi-agent systems and got lost with all the different implementation options? MCP (x)or A2A?! - Heikoโs and Dr. Sokratis Kartakis (any way to mention you in here just by your fโฆ
LLMs for the rescue?! Or are we actually building Compound AI Systems?
LLMs for the rescue?! Or are we actually building Compound AI Systems? LLMs rule the world, right?! - Only thing what matters is using the most powerful LLM available and everything falls in place. Loโฆ
2024
๐ Building your own RAG system is like deciding to build your own email server in 2024. Sure, you c
๐ โBuilding your own RAG system is like deciding to build your own email server in 2024. Sure, you could do it. But why would you want to?โ - Alden Do Rosario in his article โDear IT Departments, Pleaโฆ
โ Successful Building a GenAI use cases just requires the latest and greatest fr
โ Successful Building a GenAI use cases just requires the latest and greatest frontier model, right?! ๐ In my conversations with customers I often realize that the choice of the best & shiniest model โฆ
๐ Bagrat Ter-Akopyan, Carmen Heger and team. Congrats and thanks for the nice technical write-up.
๐ Bagrat Ter-Akopyan, Carmen Heger and team. Congrats and thanks for the nice technical write-up. Itโs amazing to see how you continue to innovate on behalf of your customers. I love the outcome. ๐ Whโฆ
How do you choose the foundation model for your Generative AI App โ like your car?
How do you choose the foundation model for your Generative AI App โ like your car? Just published a new blog post on medium: ๐ How do you choose the right foundation model for your Generative AI app? โฆ
Travelling into a long weekend and looking back to this weekโs family reunion ev
Travelling into a long weekend and looking back to this weekโs family reunion event aka #AWSSummit Berlin. To be honest, Iโm a little exhausted but energised at the same time after meeting so many cusโฆ
2023
Just in time for Santa, great book about about the role of an Solution Architect(SA) in the future.
Just in time for Santa, great book about about the role of an Solution Architect(SA) in the future. GenAI is evolving many roles - SA role is not an exception. Very interesting read, which a lot of inโฆ
2022
Contextual targeting is gaining relevance in a post-3rd party cookie area as a means for publishers
Contextual targeting is gaining relevance in a post-3rd party cookie area as a means for publishers to monetise content and for advertisers to reach their audiences. The progress in ML and the availabโฆ
2021
[German] Is your workload well-architected? Curious to understand the benefits of an architecture re
[German] Is your workload well-architected? Curious to understand the benefits of an architecture review based on best practices and to figure out how to successfully execute reviews? Tufan รzduman anโฆ
Curious how the Twelve-Factor App methodology can be mapped to #InfrastructureAsCode on #AWS? My col
Curious how the Twelve-Factor App methodology can be mapped to #InfrastructureAsCode on #AWS? My colleague Wladi โ๏ธ Mitzel and I are. Give it a read - heads-up: potential extra challenge: article is iโฆ
2020
Great work, Benedikt Stemmildt and team. Pleasure to work with you guys: Die Migration der gesamten
Great work, Benedikt Stemmildt and team. Pleasure to work with you guys: โDie Migration der gesamten Onlineplattform in die AWS Cloud ermรถglichte dem Unternehmen, sein Geschรคft weiter zu skalieren undโฆ
2019
Very good read. While I don't like the term lock-in cost, it's really a cost of change, this is a ve
Very good read. While I donโt like the term lock-in cost, itโs really a cost of change, this is a very valid point of view: Donโt look at migration cost in isolation but consider opportunity gain at tโฆ