The First Inch: How to Actually Start Using AI
At EM’s latest Lunch & Learn, we welcomed community member Aaron Plaat, founder of TripleSkinny and a digital strategist with 20+ years in marketing, technology, and web development. Aaron now runs about 95% of his work through AI and has trimmed his workweek to under 10 hours. Read on to see how.
New to AI? Take It on Dates to Start the Relationship
A few years ago, when Aaron first started using ChatGPT, he made it a weekly ritual. He’d bring his phone to his favorite restaurant, use voice dictation, and simply talk to the AI for an hour or so.
Rather than using AI to be productive, Aaron let the tool get to know him, and at the same time, discovered what it could actually do. That’s still his advice for anyone starting out. Open a new chat and ask your AI to interview you about who you are, what you do, and how you work before you ask it to produce anything.
Three Projects That Bought Back His Time
Aaron pointed to three specific jobs that used to eat entire weeks and now take a fraction of the time:
- A client proposal. Built by taking a screenshot of a client’s email with the project requirements and entering it straight into his AI chat. He iterates from there, rather than drafting from scratch.
- A PDF product sheet rebuilt as an interactive webpage. A project that would normally run 15 to 20 hours of design work took him under 30 minutes.
- A 10,000-line WordPress plugin. Rewritten end-to-end by Claude after his dev team needed to hit a fast-approaching deadline. It worked on the first upload.
Pick a Platform, Then Match the Model to the Task
There are plenty of options (e.g., ChatGPT, Gemini, Claude, Copilot), but the real gains come from committing to one and learning it well. He moved to Claude for a simple reason. Rather than just spitting out an answer, it asks clarifying questions that sharpen the output.
One practical note he stressed: match the model to the task. Claude’s model tiers (Haiku, Sonnet, Fable) and thinking levels all trade speed for power, and defaulting to the strongest option for every prompt is, as he put it, “like cutting your grass with a chainsaw.” Save the heavier models for genuinely complex jobs; use lightweight ones for quick copy and taglines.
The C.A.R.E. Framework
Vague prompts get vague results. Aaron’s shorthand for a better one:
- Context — Who you are and what you’re working on.
- Action — The specific thing you want the AI to do.
- Rules — Any guardrails: citations required, accessibility standards, brand tone.
- Example — Reference material that shows the AI what “good” looks like.
He also pointed out that AI output rarely lands on the first try, especially for anything visual. Refine one element at a time rather than asking for a finished product in a single pass.
Give It a Role, and Your Voice
Two more refinements worth stealing:
- Role-based prompting. Assign the AI a role before you start (e.g., “Take the role of a copywriter who writes how-to content for B2B SaaS”) so its output comes from that lens. Aaron uses separate Claude projects for different clients, each with its own assigned role.
- Tonality. If AI output “sounds too AI,” it’s usually because the model hasn’t incorporated your voice. Point it to your blog, your emails, or any writing sample and ask it to study your tonality before it drafts anything.
Where AI Is Going
Where are things headed? Rather than just chatting with AI, Claude Cowork, and the Claude for Chrome extension can take actions on your behalf inside a browser tab rather. Aaron’s live example: using Claude for Chrome to log into a client’s WP Engine dashboard and update plugins across production, staging, and development environments, then verify nothing broke. He’s also used it to research potential marketing partners and build out a contact spreadsheet in real time.
Higher Expertise, Now With AI Fluency Built In
Aaron put the urgency to start using AI plainly: “AI is advancing every single day. And the people that spend one week learning how to use AI will not just be one week ahead of people who pushed it off until later. They will be light years ahead.” The encouraging part is that getting started doesn’t take a technical background or a big time commitment, just a first real conversation with the AI tool.
That head start is exactly what we look for when we match consultants to clients. Every consultant on our bench brings deep marketing expertise, and a growing number bring exactly this kind of AI fluency with it. If you’re hiring for that combination, let’s talk.
