Your AI Partner Shapes the Transformation You Get
Choosing an AI partner is a leadership decision. It influences which problems receive attention, which assumptions get challenged and how the organisation approaches change.
Invent & Simplify
Essays on enterprise transformation, cloud, and building with AI.
Choosing an AI partner is a leadership decision. It influences which problems receive attention, which assumptions get challenged and how the organisation approaches change.
The studies claiming 10–30% productivity gains from AI coding tools are not wrong — they are just measuring the wrong thing. The real story is about what developers now attempt, not how fast they complete what they were already doing.
MCP's 2026-07-28 revision removes sessions from the protocol — but the state doesn't disappear, it moves through the client as an integrity-protected blob you now have to defend. What actually changed, what the coverage got wrong, and which kind of server you need to know you're building.
KAIROS is a personal AI operating system built on a Neo4j knowledge graph, MCP integrations, and agents on Amazon Bedrock AgentCore. Here's the stack, the design decisions, and what broke along the way.
Deploying AI tools shouldn’t require a DevOps degree. This guide shows developers how to deploy custom tools built with the Model Context Protocol (MCP) to Amazon Bedrock AgentCore Runtime in under 3 minutes. You’ll learn how to set up OAuth2 authentication with Amazon Cognito, launch a fully scalable HTTPS endpoint, and monitor performance—all without managing servers, containers, or scaling logic. Whether you’re building your first tool or scaling a fleet of agents, this is the fastest way to go from local script to production-ready AI infrastructure.
A grounded look at why AI feels so noisy, the seven dynamics behind hype fatigue, and what will last: products that are useful, durable and hard to copy.
Werner Vogels' six lessons on managing complexity from re:Invent 2024, from evolvability to cells and automation, plus how Canva and Too Good To Go scaled.
Essential best practices for building production-ready infrastructure with AWS CDK.
Why decisions go wrong through ignorance or arrogance, the seven traits behind bad judgement, and the values, reflection and empathy that lead to sound ones.
How to persevere at work: build a vision from your own passion, stay focused on details, work hard without burning out, and adopt seven practical habits.
Invent & Simplify
Essays on enterprise transformation, cloud, and building with AI.
Choosing an AI partner is a leadership decision. It influences which problems receive attention, which assumptions get challenged and how the organisation approaches change.
The studies claiming 10–30% productivity gains from AI coding tools are not wrong — they are just measuring the wrong thing. The real story is about what developers now attempt, not how fast they complete what they were already doing.
MCP's 2026-07-28 revision removes sessions from the protocol — but the state doesn't disappear, it moves through the client as an integrity-protected blob you now have to defend. What actually changed, what the coverage got wrong, and which kind of server you need to know you're building.
KAIROS is a personal AI operating system built on a Neo4j knowledge graph, MCP integrations, and agents on Amazon Bedrock AgentCore. Here's the stack, the design decisions, and what broke along the way.
Deploying AI tools shouldn’t require a DevOps degree. This guide shows developers how to deploy custom tools built with the Model Context Protocol (MCP) to Amazon Bedrock AgentCore Runtime in under 3 minutes. You’ll learn how to set up OAuth2 authentication with Amazon Cognito, launch a fully scalable HTTPS endpoint, and monitor performance—all without managing servers, containers, or scaling logic. Whether you’re building your first tool or scaling a fleet of agents, this is the fastest way to go from local script to production-ready AI infrastructure.
A grounded look at why AI feels so noisy, the seven dynamics behind hype fatigue, and what will last: products that are useful, durable and hard to copy.
Werner Vogels' six lessons on managing complexity from re:Invent 2024, from evolvability to cells and automation, plus how Canva and Too Good To Go scaled.
Essential best practices for building production-ready infrastructure with AWS CDK.
Why decisions go wrong through ignorance or arrogance, the seven traits behind bad judgement, and the values, reflection and empathy that lead to sound ones.
How to persevere at work: build a vision from your own passion, stay focused on details, work hard without burning out, and adopt seven practical habits.