Second Opinion · Content

AI runs the engine. It cannot be the story.

Two publishing projects built here run largely on AI. Neither one would be worth following if AI were the voice.

There are two ways to use AI for content. One is to have it write the posts. The other is to have it do everything around the posts, so a real person has time to make them.

The first is cheap, and everybody can tell. The second is where the advantage actually is. Two projects built here show the difference.

Gulf Coast Bite: the machine handles the repetition

Gulf Coast Bite publishes fishing conditions for six waters from Pensacola to Dauphin Island. Every morning, a branded conditions card is built from live NOAA tide and weather data and posted to Instagram and Facebook on schedule, with a retry if a post fails. Every Monday, subscribers get an email outlook for the water they fish. Nobody assembles any of it by hand.

What the system does not do is invent anything. Every number on every card comes from a real feed. The headlines are generated from that data, never written for effect. The captions were written once by a person and stay fixed, because a fishing report that guesses wrong loses an angler for good.

And the posts that give the account its voice are not the automated ones. They are real photos from real days on the water: a cooler of live shrimp, an osprey nest on Perdido Key, a Pensacola produce market that has been a local tradition since 1938. The automation makes room for those posts. It does not make them.

Gordy Goes North: the machine stays out of the frame

Gordy Goes North is a road journal. A traveler, two dogs and an RV head north from Atlanta toward Acadia, stopping at local animal shelters along the way to spotlight dogs still waiting for a home.

AI built the site and the tools behind it: the journal, the route updates and the shelter spotlights, all managed from a simple editor instead of a developer. For someone driving all day, that is real time back.

But the project runs on a hard rule. Every photo of the traveler, the dogs, the RV and the shelters is real. AI never generates a stand-in, however convenient it would be. People who follow a woman and two dogs across the country are following them, and the first fake photo would end that. Even the timing is a human call: locations are published after the fact, never live.

AI should make it easier to be real. The moment it starts standing in for real, the audience can tell.

Where the line sits

  • Hand AI the repetition. Formatting, resizing, scheduling, publishing, reporting and retries. This is the work that quietly eats a week.
  • Feed it real data. AI working from live numbers is an asset. AI filling a gap with a guess is a liability with your name on it.
  • Keep the photos real. A generated image of your shop, your team or your product is a promise the customer will check in person.
  • Keep the voice human. The opinion, the story and the reason anyone should care come from a person.

For a small business, this is the whole opportunity. A large brand can pay a content team. A small one can run a system that does the team’s busywork, so the owner can spend ten minutes on the part nobody else can do: showing up as themselves.

Content systems like these are part of AI-first marketing work, built around a voice that stays the business’s own.
← Back to Second Opinion