Online coaching removes observation from the programming equation. You can't watch the movement, you can't read the room, and you're making programming decisions on the basis of what the client reports rather than what you see. That changes what good programming looks like — and it changes how an AI tool should build it.
The trainer profile
The trainer in this example runs a predominantly online coaching practice with 28 active clients. Their programming philosophy for online clients emphasizes autoregulation — RPE and RIR-based loading rather than fixed percentage prescriptions, because they can't observe when a client is underloading or grinding through sets that should have been stopped. They use check-in systems to collect session feedback and adjust programs weekly based on client-reported data. Their programs include explicit movement notes and video cue reminders for primary compound lifts because form coaching happens asynchronously. They use six-week training blocks.
The client profile
The client is a 29-year-old man, intermediate training age, training four days per week in a commercial gym. He has been online coaching with this trainer for four months. His session history shows consistent attendance (average 3.7 sessions per week over sixteen weeks), reliable RPE reporting, and steady progress on his primary lifts. He has a previous history of lower back sensitivity associated with high-rep deadlift work — managed successfully in his current block by keeping deadlift sets at five reps or fewer. He is moving into a new training block with a primary goal of squat and deadlift strength development.
The generated plan: key features
The AI generates a four-day upper-lower split for this client's new block. All primary compound movements are prescribed with RPE targets rather than fixed loads. The session notes include one coaching cue per primary lift — a direct reflection of the trainer's stated preference for including form reminders in online client programs. The lower back management approach from the previous block (five reps or fewer on deadlift variations) is carried forward into the new block's loading structure.
The AI also generates a weekly check-in prompt alongside the program — a short set of questions the client answers after each training week covering RPE accuracy, any movement discomfort, and subjective recovery quality. This is built into the trainer's profile as a standard component of their online coaching workflow, and the AI includes it as part of the block deliverable rather than requiring the trainer to add it separately.
How session history shaped the new block
The sixteen-week session history for this client influences several decisions in the new block. The consistent attendance pattern supports programming a four-day split rather than building in flexible scheduling accommodations. The RPE reporting accuracy — tracked in his session history — indicates that his self-reported effort is reliable, which informs how confidently the AI can program autoregulation-based loading. The steady linear progression on primary lifts over the previous months suggests the client is not yet at a stage where linear programming should be replaced by more complex periodization — and the new block reflects that, applying a simple wave-loading structure rather than introducing undulating periodization prematurely.
What changes in the online context
The biggest programming difference in the online context is the emphasis on autoregulation and built-in feedback mechanisms. The trainer cannot adjust in real time, so the program has to include the tools that enable client-side adjustment — RPE ranges rather than fixed percentages, explicit notes about when to reduce load, and a structured feedback loop that gives the trainer the data to make weekly adjustments.
An AI tool that understands online coaching as a context — not just as training at a distance — generates programs that account for these needs rather than producing standard percentage-based programs that don't translate well to unsupervised environments.