Using AI to draft programs

This article covers using the AI assistant for program-level work: drafting full programs, generating phase outlines, suggesting exercises, and the realistic limits of what AI can do at the program level today.

The honest framing first: AI is genuinely useful for program drafting, but it's not yet capable of one-shotting a full periodized program from a single prompt and having it actually be good. This article explains why, and where AI does help.

What AI can do well at the program level

Drafting a phase outline. Ask the AI for a 4-week hypertrophy phase structure for an intermediate client, and it can give you a reasonable outline: how many sessions per week, what each session emphasizes, what the progression looks like across the 4 weeks. This is a useful starting point.

Drafting individual workouts within a program. Same as the workout-level use case. Ask the AI to draft a specific workout (e.g., "Build me a Day 1 lower body session for a beginner client"), and it can produce a workable draft.

Suggesting accessories and substitutions. If you've structured the main work and want help filling in accessory work, AI is good at this.

Filling in gaps. If you've built 80% of a program and want help finishing specific weeks or sessions, AI can do that well.

Generating program descriptions and notes. The client-facing description, the phase intent notes, the workout-level cues. AI can draft these in your voice if you give it enough context.

What AI can't do well at the program level

One-prompt full program generation. Asking AI to "build me a 12-week strength program for my client" and expecting a usable output is too much to ask today. The amount of context required (your specific client, their training history, their goals, their equipment, their preferences, your coaching philosophy, your library of exercises) is enormous, and even with all of that, the AI's output will need significant review and revision.

Periodization decisions across the full program. How to structure intensity across 12 weeks, when to deload, how to peak for a goal, how to balance volume across the program. These are coaching decisions that benefit from your judgment more than from AI generation.

Long-term progression planning. Knowing when to progress a client from one variation to a harder one, when to back off, when to add complexity. AI can suggest, but the call is yours.

This isn't a knock on AI. It's a knock on the realistic limits of what's possible given the context required. We're not pretending otherwise. As the platform improves and as we find better ways to give AI context about your specific coaching, this gap will close. It's not closed today.

Practical AI program workflows

Two workflows that work well in practice:

Workflow 1: AI for outlines, you for the build

Ask the AI to draft a phase outline (week-by-week structure, session emphasis, progression plan).
Review the outline. Adjust based on your client and your judgment.
Build the actual workouts manually, or use AI to draft individual workouts inside the outlined structure.
Review the full program before assigning.

This keeps the high-level structure under your control while using AI to speed up the actual session-building.

Workflow 2: AI for the first draft, you for the refinement

Ask the AI for a full first-draft program (all weeks, all sessions, all exercises, all prescriptions).
Treat this as raw material, not a final product.
Go through the program week by week. Replace generic exercises with ones from your library. Replace generic prescriptions with your saved ones. Adjust the structure where the AI's choices don't match your coaching.
Review the result holistically before assigning.

This is faster than building from scratch but still requires real review time. Don't shortcut the review step.

Giving AI better context

The more context you give the AI about your client, your coaching style, and your library, the better the output. A few things that meaningfully improve AI program drafts:

Tell the AI about the client. Age, training history, goals, current state, any limitations. You can also tag them or open AI from the clients dashboard and it will automatically know the context
Tell the AI about your style. Specific cues you use, prescriptions you favor, exercises you reach for.
Reference your library. "Use exercises from my library, especially the …“ tagging it with @ sign
Be specific about the program intent. Not "build a strength program" but "build a 4-week strength block for a client peaking for a specific lift, working out 4 days a week, with limited home equipment."

The AI works dramatically better with specifics than with generic prompts.

When to skip AI entirely

For some programs, AI is more friction than help. Skip AI when:

You're building something genuinely custom for a specific client where the AI would just produce something generic.
You've already got a clear plan in your head and the AI would slow you down.
You're cloning and adapting an existing program (faster to clone than to re-prompt).

AI is a tool. It's not the answer to every program-building task.

What's next

That's the Programs lesson, and that's the end of the Coaching Library module. The next module is Programming, which is more about workflows and approaches to building programs over time, less about features.

All Entries

A

B

C

E

F

H

P

S

T

U

W

Jump to letter...