Short answer: AI touches nearly everything we produce at Loveland Brand Design. That should concern you a little. It is also exactly why we built a process around it instead of a promise.
What actually happens before AI touches anything
Every piece starts with a real input: a client's actual business, actual numbers, actual history, actual voice. Part of that input is a real, current picture of who the piece is actually for, not a generic audience assumption written once and left alone. That picture gets checked against real behavior on a set schedule, and any part of it that stops predicting how people actually respond gets revised. AI drafts from that starting point. It never invents one. If we do not have a specific detail to give the model, we go get it before we write a single line.
The five-pattern check, every time
The first piece in this series named five patterns that define slop: generic phrasing, unattributed claims, no named human, repeated structure, no verifiable detail. Every draft that reaches a client, or reaches this blog, gets checked against those same five patterns before it ships, the same standard every time, not a judgment call that shifts by mood or deadline. When a prompt or a process starts producing weaker output, we roll it back to the version that worked rather than patch around the symptom.
What never gets outsourced to a model
A model can suggest structure. It can suggest phrasing. Verification and specificity still come from a person who did the actual work of finding them, every time, because a model cannot confirm a statistic is real and cannot know that a client's real differentiator is the fact they hand deliver every order within their own county. Every piece we publish can be traced back to what fed it: which draft, which check, which person signed off. AI speeds up the parts of the process that do not require judgment. The parts that do require judgment stay with a person.
We built this series the same way
Every post here ran through the same check. Statistics carry a source you can click. Claims about a real study or a real update trace back to where they came from. We hold every piece of content we produce to that standard. This topic simply makes the standard easiest to prove, because you can check our work against our own words in the same series.
The invitation
Ask any vendor claiming to write honest content exactly where their human checkpoint sits. A specific answer means a real process. We built ours around a filter we are willing to name publicly. Anything vaguer than that is worth noticing.