The Prompt Is the Problem: Why Most Coaches Get Garbage From AI

Every coach who's tried AI and walked away frustrated has a story that sounds roughly the same.

They typed something in. They got something generic back. They decided AI wasn't useful for serious coaching work and returned to doing everything manually.

The diagnosis they made was wrong.

The problem wasn't the tool. The problem was the instruction.

AI slop is your own sloppy thinking returned to you at scale.

That's not a criticism. It's a mechanism. Language models are prediction engines; they generate output based on patterns in the input. A vague input produces a vague output. Every time. Without exception.

"Write me a training program for a basketball player," tells the model almost nothing. Position. Training age. Time of season. Available equipment. Physical limitations. Session duration. Performance goals. None of that is in the prompt. So the model fills the gaps with averages, and averages are useless in a performance context where specificity is everything.

The output looks like a training program. It has sets and reps and exercise names. It just has nothing to do with your actual athlete in your actual situation.

That's not a hallucination. That's what happens when you give a capable tool an incomplete task.

The fix isn't a better AI. It's a better prompt.

Structured prompting is a communication skill, and experienced coaches already have it. You give task-specific instructions to interns, assistants, and medical staff every day. You know how to explain what you need, what the constraints are, and what good output looks like.

Apply that same precision to AI, and the output changes completely.

A structured prompt gives the model five things: a role, a context, a set of constraints, an output format, and, when relevant, an example of what good looks like. That's it. No coding required. No technical background needed.

The difference between a vague prompt and a structured one isn't complexity. It's specificity.

Here's what that looks like in practice.

Vague: "Write a return-to-play progression for a hamstring strain."

Structured: "You are a strength and conditioning coach working with a professional basketball player returning from a grade 2 proximal hamstring strain. He is 18 days post-injury, cleared for straight-line jogging, and has no pain at rest. The team has 14 days until his target return-to-play date. Build a daily load progression for the next 14 days. Include session goals, volume parameters, and criteria for advancing each phase. Format as a day-by-day table."

The second prompt takes ninety seconds to write. The output it produces takes ninety seconds to evaluate and adjust (always evaluate and adjust)

That's the exchange. A small investment in precision on the front end. A usable first draft on the back end.

The coaches who dismiss AI after one bad output are the ones who handed a sharp tool to someone without a briefing and were surprised when the result was wrong.

This isn't unique to AI. It's how expertise works. The tool performs according to the quality of the instruction.

What changes when you start prompting with structure is that you also start thinking more clearly about your own workflows. When you have to articulate the role, context, constraints, and output format, you realize how often those elements were fuzzy even in your own head.

That's a secondary benefit the industry doesn't talk about enough. Better prompts reveal unclear thinking. And clearer thinking produces better coaching decisions, with or without AI.

The technology is not the barrier. It never was.

The barrier is the same one it's always been: precision of thought, translated into precision of instruction.

AI just made that barrier visible.

————
The full prompting framework for S&C workflows — built specifically for coaches, not developers — is in Prompting for Performance: The AI Playbook for Strength & Conditioning Coaches. Available on Amazon for $9.99.