// on building for people
I don't build software for the sake of building software. I build it because somewhere on the other end, there's a person whose day gets a little better. That's the metric that matters.
// the approach
Before writing a line of code, I ask: who is this for? What are they trying to accomplish? What frustrates them about how it works today? The answers shape everything — the architecture, the interface, the priorities. Technology that doesn't serve people is just expensive complexity.
A customer service agent handling a pension claim shouldn't need to understand DynamoDB, Lambda, or BPMN. They should see a clean task, clear information, and obvious next steps. The complexity is real — orchestrated workflows, event-driven processing, AI-powered decisions — but it lives under the surface. The user never touches it.
A process that takes two weeks by mail can take two hours digitally. A customer who waits three days for a response can get an answer in seconds through an AI agent. Every second of unnecessary waiting is a failure of design. When you optimize for the customer's time, everything else follows.
// in practice
Customer service agents at a pension fund handle dozens of cases daily. The tasklist I built prioritizes what matters: urgent cases surface first, all context is on one screen, and the next action is always obvious. No switching between systems, no hunting for information. The tool gets out of the way so the human can focus on the person in front of them.
When a member requests a change to their pension, they shouldn't wonder if it went through. The process is orchestrated end-to-end with BPMN — validation, compliance checks, approval routing, and confirmation — all happening in the background. The member sees a clean status update. The system handles the rest.
Most chatbots are fancy search bars that make you rephrase your question three times. The agents I build are different. They understand context, pull real data, and take real actions. An email agent that drafts a proper response. A voice agent that resolves the issue on the call. The bar is: did the customer's problem get solved?
A legacy system with no API is a system that only insiders can use. The AWS integration layers I build don't just connect systems — they unlock data and capabilities that were previously buried. When customer service can see everything in one place, the member on the other end gets a faster, more accurate answer.
// convictions
The best technology disappears. If the user notices the tech, you've failed.
Measure what matters to customers, not what's easy to track.
Every workaround a user invents is a bug in your design.
Fast feedback loops beat perfect planning.
Automation should create time for human connection, not replace it.
Want to build something that makes a real difference for people?