❧ Butler
Private pilot · limited seats

For consultants & client-service agencies

Never miss a client reply.

On-brand drafts, waiting in your Gmail before you've had coffee. Butler watches for replies from your clients — not your whole inbox — and leaves a reply draft in your voice, grounded in that client's actual project state. You read, tweak, send. It can't send anything itself.

Free for design partners · 2-minute Gmail connect

How it works

Step one

Connect Gmail & pick your clients

OAuth in two clicks. Butler only acts on mail from the client list you give it — everything else is invisible to it.

Step two

It learns your voice — from you

A one-time read of your own sent replies builds a private voice profile: your greetings, your sign-offs, your way of saying no. Nothing is trained; nothing is shared.

Step three

Drafts appear, you decide

Client replies get triaged; the ones that need answers get a draft in your Drafts folder, labeled AI/Drafted. A morning digest sums up the rest.

It knows the account, not just the thread

Generic email AI drafts from the conversation. Butler drafts from the client's actual state — that's the difference between "sounds right" and "is right."

Live SEO numbers

Your Ahrefs, in your replies

Connect your own Ahrefs key and drafts cite what's true today — domain rating movement, new referring domains, keyword jumps — per client, refreshed daily.

Your documents

Grounded in your own files

Hand Butler a client's proposal or monthly report from Google Drive (you pick the files — it can't browse) and drafts stop being generic: they reference what you actually promised and delivered.

Your facts

Client notes, remembered

Retainer terms, touchy topics, who hates jargon — per-client notes feed every draft. And Butler quietly learns from what you actually send, so your voice profile sharpens month over month.

Built to be trusted with a client inbox

"Connect your client email to an AI" is a big ask. These aren't settings — they're what the product is.

Draft-only, by design There is no send button in our code.
Minimal permissions Our access cannot archive or delete your mail.
We never store your email Bodies are processed in the moment and discarded; we keep only message IDs and labels.
No AI training on your data Your mail tunes your private voice profile and nothing else — never a shared model.
Leave in one click Disconnect revokes access and erases stored metadata.
Pilot-phase honesty — during the pilot you'll see a "Google hasn't verified this app" screen at connect time. Google's verification for exactly these permissions is in progress; that warning is the system working as intended. We ask you to read it and decide.

Pricing

Solo

$79 /mo

One mailbox, unlimited clients.

Agency

$59 /seat/mo

3+ seats. Shared client rosters and project factsheets across the team.

Design partners: free pilot, then 50% off for 6 months. (Prices are the P0 hypothesis — validate in discovery calls before printing.)

Questions people actually ask

Why does Google's consent screen mention "send"?
Google has no drafts-without-send permission; the smallest scope that can create drafts includes the word. Our code has no send path — that's the product's whole premise, and it's contractual in our terms.
What AI reads my mail?
Anthropic's Claude, via their commercial API, which doesn't train on customer content. Your mail never becomes anyone's training data.
What does it integrate with?
The pilot ships with Ahrefs (your own API key) and Google Drive (files you pick), plus per-client notes. Sheets, Search Console, GA4 and custom-app connections follow based on what pilot partners ask for first.
Does it reply to everything?
No. Only mail from your client roster is even looked at, and only the messages that actually need an answer get drafts.