Email

Email is one of the highest-value integrations in Jootle, because it lets your AI participate in threads with people who don’t use Jootle, and lets you forward, CC, and dictate from places where you’d otherwise lose the context.

There are two roles your email account plays once it’s connected:

Inbound. Your AI can read threads you forward, CC it on, or ask it to monitor. It pulls context, summarizes, drafts replies, and tracks follow-ups.
Outbound. Your AI can compose and send messages as you, after approval (see Approvals and gates).

This chapter covers what you can do once it’s set up. For the setup itself, see Connect Google Workspace (or the equivalent for other providers, when those are documented).

Forwarding to your AI

The simplest pattern: forward an email to your AI’s inbound address (looks like <your-instance>@in.jootle.com) with a short note about what you want.

“Forwarding the contractor’s quote. Compare it to the other two and tell me what stands out.”

“FYI, file this thread under the kitchen project and let me know when Mike replies.”

“Summarize this in three bullets and add to the brief for tomorrow.”

The forward arrives, your AI reads it, files it appropriately, and either replies inline (if you asked for a response) or quietly tracks the thread (if you asked to monitor).

You can forward attachments too. Documents land in the project’s files. Images get OCR’d if there’s text to extract.

CC’ing your AI on a thread

Add your AI’s inbound address to the recipient list on any thread you’d like it to follow. Your AI will see every message in the thread going forward, can summarize on demand, and can draft replies.

You (in the chat with your AI): “Draft a reply to Mike on the kitchen thread. Accept the quote, push back gently on the timeline.”

Your AI: [produces a draft, gates on send]

CC’d threads are scoped: your AI sees the messages in that thread and what’s been forwarded to it. It doesn’t read your entire inbox unless you ask it to (and you’d ask via a goal or follow-up, not by default).

Inbox monitoring

When you want your AI to watch your inbox for specific patterns, you set up a goal or follow-up.

“Watch my inbox for anything from the lawyer. If something arrives, summarize it and ping me on Telegram immediately.”

“Every morning at 7am, scan the overnight inbox and tell me about anything urgent.”

“If a customer named in the support pipeline emails me, file it under their record and draft a starter reply.”

The watcher runs on your AI’s schedule. Most monitoring doesn’t fire most days; you’ll only hear from it when the trigger matches. That’s the point.

You can also stop monitoring at any time: “Stop watching for emails from the lawyer.”

Sending email as you

Once you’ve connected an email account, your AI can compose and send messages from it, after approval.

“Email Mike: ‘Quote accepted, can we start the week of June 5?’ Sign with my usual sign-off.”

Your AI drafts, you approve (or revise), and the message goes out from your actual address with you as the sender. Recipients don’t see anything indicating an AI wrote it.

You can also batch:

“Send a polite follow-up to everyone I’ve asked for a quote on the kitchen project who hasn’t replied in over a week.”

Your AI identifies the open quote threads, drafts an individualized follow-up for each, and gates them all for batch approval. You approve the lot at once or revise the ones you want different.

Drafts that wait

Sometimes you want a draft prepared but not sent.

“Draft a thank-you note to Anna for the kitchen referral but don’t send. I’ll review and send myself.”

Your AI produces the draft and saves it as an artifact. You can copy and send it from your normal email client when you’re ready, or come back later and ask your AI to send it.

Specific patterns that work well

A few patterns customers come back to:

Triaging the inbox. “Go through my unread messages from the last 24 hours, categorize them as urgent / can-wait / ignore, and tell me which deserve a reply today.”

Drafting before replying. “Read this thread and draft three possible replies: short and warm, neutral and clear, and one where I push back on the timeline.”

Following up on something you sent. “I sent the quote to Mike on Tuesday. Watch for his reply. If he hasn’t responded by Friday, draft a follow-up I can review.”

Extracting structure. “Take this email thread between me and the contractor and produce a decisions list of what we’ve agreed on, separate from a list of what’s still open.”

What it doesn’t do (yet)

A few things to set expectations:

Doesn’t archive or label aggressively. Your AI can do these on request, but it doesn’t reorganize your inbox without you asking. We don’t want a runaway labeler.
Doesn’t sift through your entire inbox on first connect. Connecting your account doesn’t trigger a backfill. Your AI starts learning from new mail and from threads you specifically reference.
Doesn’t auto-reply on its own. Even a thread your AI is monitoring won’t send a reply unless you’ve explicitly told it to.

These limits are intentional. Email is high-stakes; we’d rather be too quiet than too loud.

Stopping your AI from being too involved

If at any point you want your AI to back off:

“Stop watching the lawyer thread.” Removes the follow-up.
“Don’t send any emails for the rest of the day.” Pauses outbound until you say otherwise.
“Disconnect this email account.” Goes to Integrations and removes the account entirely. (Your data isn’t deleted; you can reconnect.)

The reverse moves are equally easy. You can scale up and down with the rhythm of your work.

A small note on signatures

Your AI will use whatever signature you’ve set in your email client by default. If you’d like a different signature on AI-drafted mail (some customers prefer “Drafted with Jootle assistance” for transparency), you can configure it in Settings. Most customers don’t bother; the messages are yours either way.