We were early six times.
Twenty-five years of arriving before the category had a name. Here is each one and, so you can place it, the phone you were carrying at the time.
In the beginning there was HTML, and it was dull. So we started building experiences that moved, responded and played. The industry filed it under decoration for years, by which time a browser had learned to behave like an application.

DIRECTV was selling subscriptions by the million across Latin America. We built the infrastructure behind those sales and measured the results, across 12 countries and ten years. That work is what made us a Google Analytics Certified Partner.

Everyone agreed mobile mattered, and almost everyone still drew the desktop version first. We inverted it, and picked up Angular while it was still a curiosity: single-page apps, the whole thing running in the browser, the server reduced to an API. Both bets took years to become the default.

We built our first headless CMS by serving custom JSON out of WordPress, a good while before anyone in the room had heard of Contentful. Once content traveled over an API the hard part moved underneath it: what a "customer" actually is, and how to get one out of systems that were never designed to share.

We put LLMs into our own process within weeks of launch, well before a client thought to ask for it. The direction was already clear: anything you could describe precisely was about to get an order of magnitude cheaper. Three years of using it on ourselves is what made the next step possible.

withTimeout.withRetry, matching the signature. Two things I was unsure about: whether 429 counts as retryable, and jitter. I left jitter out.withTimeout.withRetry, matching the signature. Two things I was unsure about: whether 429 counts as retryable, and jitter. I left jitter out.Same pattern as the other five: we rebuilt how we work before we sold a minute of it to anyone. Agents execute, people own every gate that matters, and an auditor went through the whole thing. It cost us a year and it broke a few things. It was worth it.

Twenty-five years.
Six times early.
Every one of those bets is still running somewhere inside the work we do today. The current one has a shape you can look at.
See how a pod works