AI CONTENT SYSTEMS
Milking copy out of ChatGPT is easy. Getting an empty chat to nail a client’s voice and meet agency standards isn’t. So, I developed reusable workflows with context baked in, cutting repetitive work without handing over the creative decisions.
THE OBJECTIVE
As more people started using AI, the same issues surfaced across departments: inconsistent output and too much back and forth. Everyone was using it differently for different jobs, from writing social and newsletter copy to preparing for client meetings. The common thread was the amount of trial and error between opening ChatGPT and getting something useful back. That was my focus as I shifted from lead copywriter to AI strategist: to build systems that improved quality and capacity, without turning AI into a crutch.
THE EXECUTION
I started with my own team, asking web writers to walk me through exactly how they were using ChatGPT. Then, I tested each approach and turned what worked into a shared prompt library, refining it as the models and our understanding of them changed.
I also pulled Lifted Logic’s SEO, readability, grammar, and formatting standards into one AI-readable guide, building it into the setup process for every project. Writers loaded the client’s voice survey, service details, sitemap, page outlines, and internal standards into a client-specific ChatGPT project with project-only memory. That gave the model the same source material the writer was using, without letting one client’s context wander into another.
Then, I moved on to ironing out the social and newsletter copy workflows. Once again, I started with prompt libraries covering all the unglamorous parts like formatting rules, CTA logic, and how to sound like a human. For the media team, I created a separate process that used the client’s voice survey, service details, and video requirements to write a shot-by-shot script for every homepage video.
Once the social and newsletter workflows held up, I turned them into dedicated agents for ongoing clients. Each one had its own voice rules, service facts, approved examples, platform logic, CTAs, and hard “don’ts.” I built and tested every agent, audited project setups, and trained strategists to use and refine them.