I haven't written a log since August. Not because there was nothing to say - the opposite. The last three months have been heads-down building, and the building has been the writing. But something clicked today that I want on the record with a date on it, because it reframed the entire year for me.
It dawned on me what it must have felt like to be an entrepreneur in the early days of the internet. Not the version we tell now, with hindsight doing all the work. The version from the inside, in 1996, when you could see the thing clearly and almost nobody around you could.
The View From 1996
Here is what I think that felt like. You were not smarter than the people dismissing it. You simply took the new thing seriously while they were busy explaining why it didn't count. In 1995 a well-known astronomer wrote a piece for Newsweek titled "The Internet? Bah!" arguing that no online database would ever replace the daily newspaper and nobody would buy books over a network. A Nobel-winning economist later predicted the internet's effect on the economy would be no greater than the fax machine's. These were not fools. They were serious people looking at an early, clumsy version of a thing and judging it by what it was instead of what it was becoming.
Meanwhile, somewhere, a person with a modem and no particular credentials was registering a domain name and building something that would quietly compound for twenty years. Not because they were visionaries. Because they were paying attention, and the price of admission was attention.
I wrote in April about teaching myself HTML in 1997 at sixteen, convinced everyone would learn it. They didn't. And HawaiiGuide itself was born in 2002, in the smoking crater of the dot-com bust, when the consensus had swung hard the other way and the internet was supposedly a party that had already ended. It wasn't. It had barely started. The people who kept building through the "it's over" phase were the ones who were there when it actually arrived.
I didn't understand that at twenty-one. I just liked building things about Hawaii. But I'm living the second version of that story now, and this time I can see it while it's happening.
The Same Wind
In February I wrote The Click That Disappeared, about watching HawaiiGuide's traffic fall by more than half while our rankings improved. AI Overviews ate the click. That post was honest about the fear, and the fear was justified. Every independent publisher I know is standing in that same wind.
And this is the thing I finally saw clearly today: it is the same wind. The AI that is a headwind for a publisher fighting it is a tailwind for a publisher who sets the sails. Not a different force. The same one, met differently.
Fighting it looks like paywalls, crawler blocks, lawsuits, and a steady stream of articles arguing that the machines are a bubble, a fad, a fax machine. Setting the sails looks like what I documented in August: a two-person family company directing agentic coding tools against a quarter century of Hawaii knowledge and shipping, in about 130 days, 468 milestones and 47 free traveler tools - live ocean reports, a trip planner, an itinerary checker that verifies what the chatbots write, and a data layer built for the agents themselves.
I pulled that same internal timeline again this morning. It now reads 1,030 milestones through November 7. That is 562 entries in the 96 days since the last post, and I want to be honest about what that number is and is not. I slowed down. After the launch sprint I shifted into maintenance mode, and the count kept climbing anyway, because a great deal of it is the system doing its job. Only one new tool launched in that window - the Visitor Dashboard, a public beta on October 26. About 125 entries are the daily data runs that rebuild the weather, ocean, drive-time, gas-price, and briefing feeds every morning without anyone touching them. The rest are people and machines keeping the truth current: Kīlauea episodes 54 through 56, a Kauaʻi flood recovery tracked for weeks until the parks and roads reopened, three 2027 guidebook editions, and a few hundred page refreshes.
And then there was the weather, which deserves its own paragraphs, because it is where this direction started and where it got tested. When the Kona low storms hit the islands back in March, the live-conditions system did not exist yet. We did not cover them at all. But watching them was the moment the light bulb went off. The agencies were publishing good data - the weather service, the counties, the parks, the highway crews - and it was scattered across a dozen sites with no clear place for a visitor to go. The chatbots were no help either, because a language model has real trouble telling the authoritative source from the thousand confident ones. There was an obvious job sitting there: vet all of it, use a quarter century of experience to decide what a visitor actually needs in front of them, and put it in one place.
By August that system existed, and hurricane season became its first real-world trial - Lala, Moke, Lowell, Karina, and Nolo, tracked advisory by advisory, sometimes two systems at once, with recovery guidance staying current for weeks after each one passed. It kept pace. Nothing about that coverage was possible on the March version of the site, and I have come to think it is our new thing: not more articles, but a trusted clearinghouse for what is true in the Islands today. Launches are what a sprint looks like. Maintenance is what a tailwind looks like once the sails are set: the boat keeps moving while you catch your breath, and it holds its heading when the weather turns.
HawaiiGuide has done very well this year with AI in its sails. I don't say that to gloat, and I say it carefully, because I know how many good people are fighting the headwinds right now and how exhausting that fight is. I say it because it is the data point. Same storm. Different trim.
The Word "Responsible"
I want to be precise, because the easy reading of this post is "use AI and win," and that's not the lesson. Plenty of people are using AI to flood the internet with confident garbage, and that is a headwind of its own making. It erodes exactly the trust that makes a resource worth visiting.
What has worked for us is proper and responsible use, and both words carry weight. Proper means the machine does what machines are good at - the code, the scaffolding, the thousandth variation of a data pipeline - while the ground truth, the judgment, and the standards stay human. Responsible means a verification gate in the editorial process so nothing ships on vibes, live data in place of a model's memory of 2023, and a named person who answers publicly when something is wrong. The AI amplifies the craft. It doesn't replace the accountability. The moment it does, you've traded a tailwind for a reef.
That distinction is also why I have patience for the people who are rejecting it outright. Some of them are rejecting the slop, not the tool, and on that point we agree completely. Some are tired. Some have real ethical objections I take seriously even where I land differently. What I have less patience for is the third group - the ones not rejecting it so much as declining to look. That is the Newsweek essay. That is the fax machine. That is the posture the early internet punished most, not because it was malicious but because it was comfortable.
Grateful
I used to think the early-internet founders got lucky. I don't think that anymore. Luck is being in the right place by accident. They were in the right place because they noticed where the place was, and noticing is a choice you make every day for years before it pays off.
So I'm not grateful for being right. I'm grateful I was paying attention at the moment when attention was the entire price of admission - that the sixteen-year-old with the 28.8 modem grew into someone who still looks at a clumsy early version of a thing and asks what it is becoming. I'm grateful that this time, unlike 2002, I can see the shift while I'm standing in it. And I'm grateful it is happening to a small family company in Hawaii's corner of the internet, where the whole point was always to build the real thing and give it away.
The wind is the wind. It doesn't care about your business model. The only question that was ever yours to answer is which way the sails are pointed. 🤙
