Practical AI adoption

Start with the problem.

Before choosing a tool, I want to understand the workflow, the people involved, and what would make the outcome better. That gives an experiment a purpose. It also makes it easier to decide where AI earns its place and where it doesn't.

Make experiments small.

An idea becomes easier to discuss when there's something to try. A small prototype can reveal an unclear requirement or a weak assumption before a team commits more time. The useful question is what we need to learn next, and how little we can build to learn it.

Keep judgment with people.

AI helps me explore options, challenge assumptions, and produce work. I still have to decide what matters, check the result, and own the consequences. Scope, trade-offs, and the decision to continue belong to the people doing the work.

I find AI useful in four connected ways: thinking more clearly, communicating better, learning faster, and improving what I deliver. That can mean comparing options before making a recommendation, shaping a rough idea into a clearer message, or getting an initial map of an unfamiliar topic.

The quality of the result still depends on what I bring to it. In an experiment I wrote about in July 2026, I tried making an AI avatar from a short selfie video. The voice and facial movement weren't quite right, but producing a video was becoming easier. The harder part remained having something useful to say. I see the same distinction in everyday work: faster production only helps when the thinking behind it holds up.

My archery app is a personal example. I wanted a way to record my children's practice, technique, equipment, and coaching notes. As I described in June 2026, working through flows and a prototype with AI helped narrow that broad idea to a first version focused on recording a training session. Something concrete was easier to question and change.

In July, I wrote about the other side of that speed. The first version had grown without clear enough boundaries, and I rebuilt it. With a specification, I decided the scope, non-goals, and offline-sync rules before implementation. Those decisions made the work easier to carry through.

Design needed similar clarity. Seeing two layout options side by side helped me express what I wanted more effectively than another round of verbal corrections. Across these experiments, AI helped make possibilities tangible; choosing between them remained my job. These are lessons from a personal project, with its own scale and constraints.

At Emtek, I've delivered AI training for leadership and teams across News and Moji. The focus is on recognizing useful business problems and applying AI in everyday work while keeping human judgment central.

I also joined the AI Leadership Think Tank panel at MMA SMARTIES Unplugged Indonesia on 14 April 2026, discussing practical AI adoption in media and content production.

Talks, training, and event sources →