Messenger concierge — the practitioner setup that saves six minutes per client touch.

A coach I work with logged her week before we built this. Twenty-eight client touches — quick check-ins, reschedules, supplement questions, one acute query, the usual mix. Average length under five lines. Average time to deal with each one: just over six minutes, mostly because she was switching between four places to find the last message and re-typing the same paragraph she had sent eleven other clients that month. Multiply that out and the maths is grim: nearly three hours a day on the messenger shuffle, before she opened a single client file.

Two weeks later, after wiring a reasoning chat into the messenger her clients already used and writing a one-page standing instructions block for her practice, the average dropped to about ninety seconds. Not because she replied less. Because the drafting, the cross-checking, and the follow-up scheduling now happened inside the same thread — and her job collapsed to read, edit, send.

the setup, in one paragraph

One messenger your clients already use (most picked WhatsApp; a few used Telegram; one used Signal for a specific cohort). One reasoning chat tool, added as a working contact she could forward client messages to. One standing instructions block — written once, refined for a fortnight — that told the assistant: this is the practice, this is the voice, these are the four reply templates we use, this is what we never advise on, this is the format for handing the draft back, this is when to flag something for human attention.

what the assistant actually does

1. Drafts the reply in the practitioner's voice, using one of the four templates and the relevant client context. The practitioner reads, edits, sends. Nothing leaves without a human approval gate.
2. Pulls the relevant note from the previous session (or the last three) into the draft so the practitioner is not scrolling for context.
3. Schedules the next touch — "check in Thursday morning, ask about sleep before the protocol kicks in" — without the practitioner opening a separate reminder app.
4. Flags anything outside the practice's scope (a symptom that needs a GP, a question that touches dosing, a tone shift that needs a phone call) instead of drafting through it.

where the line stops

The assistant never sends a message directly to a client. Every reply is a draft, sent into the practitioner's thread for review and one tap to forward. The standing instructions block names what is out of scope and what triggers a human escalation — those are not defaults the tool picks; they are decisions the practitioner makes once and the tool then honours. Nothing about this setup tries to replace the practitioner. It removes the typing, the scrolling, and the context-switching — and leaves the judgement entirely in the room.

what changed in her practice

Across one quarter: average touch time fell from six minutes to about ninety seconds, freeing roughly nine hours a week. Two of those hours went into more time per discovery call. One went into the Sunday letter ritual she had been postponing for a year. The rest stayed unbooked, deliberately — the point was not to take on more clients, it was to do the work she already had without the messenger tax. Client NPS at the next quarterly review was unchanged. The drafts read like her, because the standing instructions block was written from her own previous replies.

the shape of the standing instructions block

Eight short sections, none of them clever, all of them load-bearing. Voice and register. The four reply templates (with examples drawn from the practitioner's own previous replies). The explicit no-fly list — never advise on dosing, never diagnose, never extend scope beyond what the practice handles. The handover format the assistant must return drafts in (so the practitioner can read in three seconds). The escalation triggers — what tone shift, what symptom, what kind of question pulls the draft and asks for the practitioner directly. A weekly review prompt the assistant runs every Sunday so the block itself stays current. A short list of things the assistant should remember about the practice (typical session length, supplement scope, referral partners). And a one-line ethic the block opens with — the practice's posture in plain English.

if you want the full system

The free setup at /setup/ai-assistant-in-your-messenger-in-10-minutes is enough to get you started inside a coffee break — same shape, individual scope. The paid pack at /resources/messenger-assistant-setup-pack includes the practitioner extension: the full standing-instructions block, the four reply templates, the escalation triggers, the handover format, the Sunday review prompt, and the troubleshooting page for the three places this breaks the first week. Installments are available at checkout.

If you would rather have us wire it for you — your voice, your templates, your no-fly list, your escalation triggers — the Messenger Concierge Setup at /done-for-you runs a one-week build with the standing instructions block tuned over a fortnight against your actual previous replies. The case study at /case-studies/practitioner-auto-followup-from-messenger-2026-05-24 walks through the same coach above: numbers, before/after, and the small handful of things that did not work.