Everyday guide to AI prompt engineering

Aimless AI prompts lead to aimless AI answers. Fortunately, the opposite is also true.

Before asking AI to tackle a job, decide exactly what success looks like. Do you want readers to feel a certain way? Is there one aspect of the project that deserves extra attention? If so, say so.

Goshorn emphasized the importance of setting your own end goal before sending a prompt to AI. "The people that do the best with AI have a couple of things in common," he says. "Thought and a vision; they know what they want out of it."

The clearer your destination, the easier it is for AI to help you get there. You can always steer the conversation through follow-up prompts, but it's usually easier to point AI in the right direction from the start.

Artificial intelligence is a powerful tool, but it can't get inside your head without an explanation. Tell AI details about how the final product should look. Explain your preferred style and tone, length, structure and layout, welcome and unwelcome elements and any other specifications that matter.

"How do you want AI to act?" Goshorn asks before listing a few potentially helpful instructions. "Don't always agree with me. Don't use jargon. Use bullet points. Include the citations. Show the source for the citations."

Your requests may look different from Goshorn's because your goals probably aren't the same. According to him, that's exactly why specific instructions matter. Without them, AI won't know how to meet you where you're at.

The longer an AI conversation continues, the likelier the tool generally is to move away from your original intent. This process, known as context drift, occurs because each detail across multiple prompts becomes part of a larger cloud of information, and AI won't understand which specifics matter most.

You can get AI back on track by reminding it what matters most throughout the conversation. Often, though, the simplest solution is to begin a new chat without the baggage from the previous one.

Greg Goshorn Owner, AI Growth Map and Mapping Your Marketing

"If you've had these long sessions, you may feel like AI gets dumber and forgets what you started to talk about at the end because there's a context window," Goshorn says. "So, if you're doing something that's really long and involved, it makes a lot of sense to break it into sections."

Starting fresh usually removes the extra context that may be getting in the way, but it also means losing helpful background information. When breaking a project into smaller pieces, it's best to finish one milestone before opening a new conversation and carrying over only the key takeaways and guardrails.

While AI speaks with confidence, it lacks human judgment to tell when its output is true or false. Blind faith in its results could leave you with inaccurate information that harms your brand, business strategies or understanding of a situation or idea.

Trusting AI without checking its work puts your credibility on the line. Small mistakes may slip through now and then, but eventually they'll catch up with you.

Cincinnati's healthtech startups soar with state funding

March 6, 2026

Four Cincinnati-based healthtech startup ventures earned significant grants from the state of Ohio, fueled by expert guidance and resources at UC's 1819 Innovation Hub.

Guide to using AI tools for small business

September 8, 2025

AI is a powerful tool for small businesses to reduce inefficiencies, automate routine tasks and cut costs. We'll explain how entrepreneurs can leverage AI to launch startups to greater success.

1819 startup builds AI engineer to help developers code

July 23, 2025

Tembo, an AI-first engineering startup based at UC's 1819 Innovation Hub, recently rolled out a pioneering AI agent to boost software engineering throughput.