4h51 sleep, third short night. Deload drafted.
Drifting for three days. Nothing sent yet.
Recovered. Green to push.
Needs you
M. Okafor · Deload drafted
HRV down 18% on a third short night. Knee volume capped. Heavy day moved off the morning.
Nothing reaches a client until you send it.
Ask in plain language, on desktop or iOS. Harlen answers from that client’s own context model: the sessions, the notes you logged, the data underneath. Nothing reaches the client until you send it.
You
How’s Okafor’s knee holding up?
Harlen
Cleared under load since Tuesday. Volume stays capped through Sunday.
Deload drafted for Thursday. Waiting on you.
Answered from M. Okafor’s context model
Harlen reads every client the way you would, does the work, and comes back for your approval.
A six-week block in about five minutes. Coaches we work with put a block at roughly forty-five minutes by hand. Harlen already knows the goals, the injury history and the last cycle, so nothing is restated. It edits blocks you built yourself the same way, and every edit teaches it.
ChatGPT is a generalist that starts from nothing every time and cannot see your clients’ data. Harlen holds a model of every client, reads their wearables against their own baseline overnight, and learns how you coach: every edit you make teaches it, so each block comes back closer to how you would have written it.
No. It brings the evidence and the reasoning, and you make the call.
Only what you approve. Drafts queue on your side, and nothing reaches a client until you send it.