Pillar 1 · Email-in capture

Stop transcribing your inbox into your to-do list.

The task was already written — by the client who emailed you. Forward that email to your private LoopVantage address and AI pulls out the commitment, the deadline, and who's involved. You tap yes. That's the whole workflow.

Works from Gmail, Outlook, or your phone's mail app · No inbox access, ever

The mechanic

Forward. Extract. Accept.

Three steps, and only one of them is yours to do twice.

1

Forward or BCC it

Every account gets a private capture address. Forward the email with the buried ask — or BCC the address when you make a promise — from any mail client, on any device. No plugin, no extension, no new app to open.

2

AI reads it so you don't re-read it

The extractor finds the commitments inside the thread: what needs doing, who asked, who owes whom, and any deadline hinted at along the way. Each becomes a candidate card with a verb-first title — “Send the revised SOW,” never “Client stuff.”

3

One tap makes it real

Candidates land in your Task Inbox for a yes or no — Y/N on the keyboard, a tap on your phone. Nothing enters your list without your explicit accept, so your list stays yours, not the AI's.

The ask was buried on line 4 of Tuesday's email. Now it's a tracked commitment — and you never typed a word.

What the AI actually finds

Three kinds of loops hide in every thread

LoopVantage sorts each extracted item by which way the obligation points — because a promise you made and a reply you're owed deserve very different treatment.

Action required

Things you must do

Someone asked you for something — a review, an answer, a deliverable. It becomes a candidate task with the deadline they mentioned, so the ask stops living rent-free in your working memory.

Commitment made

Things you promised

“I'll get you numbers by Thursday” is a reputation on a timer. These items carry the most weight in your Do This Next ranking — because breaking an external promise costs more than a missed chore.

Waiting for

Things you're owed

Forward the ask you sent, and accepting it arms the Follow-Up Radar — the part of LoopVantage that nags until the reply arrives. The follow-up black hole closes at the moment of capture.

The Task Inbox

Every candidate shows its work.

No mystery tasks appearing in your list. Each extraction arrives as a card that tells you what kind of loop it is, how confident the AI is, and exactly which email it came from — with a deep link back to the source thread in Gmail or Outlook. Skim, tap, done in seconds.

  • Kind badge: action, commitment, or waiting-for
  • Confidence score — uncertain extractions get flagged, not hidden
  • One-click jump back to the original thread
  • Accept, edit, or reject — Y/N keys or a tap

Task Inbox — 2 candidates from forwarded mail

Commitment madeconfidence 0.93

Send the revised proposal to Marcus

Due Thursday · counterparty: Marcus (client) · open source thread

Accept (Y)EditReject (N)
Waiting for low confidence · 0.64

Chase Priya for the publisher shortlist

Expected reply by Friday · accepting arms the Radar

Accept (Y)Reject (N)

Why forwarding wins

Achieve inbox-to-task capture without handing over your inbox

Every other tool solves this by asking for full mailbox access. We think that trade is worse than the problem.

One address, every inbox

Split between a Gmail and an Outlook account — or three of each? Forwarding works identically from all of them, simultaneously. Inbox-sync tools make you pick a mailbox; your capture address doesn't care which one the email lives in.

You choose what the AI sees

Nothing scans your mailbox looking for “tasks” in your newsletters, your bank alerts, or your personal threads. The AI reads exactly the emails you forward — and only those. A curated diet also means sharper extractions.

No inbox scopes to trust us with

Sign-in with Google or Microsoft is identity-only — LoopVantage never requests permission to read your mail, so there's no permission to misuse. Read the privacy architecture for the full picture.

Want full autopilot anyway? Opt-in, read-only ambient inbox sync is on the roadmap. Forward-based capture stays forever — some things shouldn't require handing over your mailbox.

The fine print, proudly

Small details that keep capture trustworthy

Verified senders only

Your capture address accepts mail exclusively from the addresses you've confirmed with a one-time code. Anyone else trying to inject tasks into your list is shouting into the void.

Raw email purged

After extraction, the original message is deleted. Only the commitment, a short context snippet, and the subject line remain — enough to act on, nothing to hoard.

Duplicates get flagged

Forward the same thread twice and LoopVantage surfaces “possible duplicate of…” instead of silently doubling your list — or silently merging things it shouldn't.

Deep links to the source

Every task keeps a one-click link back to the original thread in Gmail or Outlook, so the full context is one tap away when you actually do the work.

Low confidence, clearly labeled

When the AI isn't sure, it says so. Uncertain extractions arrive flagged for a closer look instead of masquerading as facts — and truly weak guesses never reach you at all.

Radar hand-off built in

Accepted waiting-for items flow straight into the Follow-Up Radar with counterparty and expected-reply date pre-filled. Capture once, get chased automatically.

FAQ

Fair questions about capture

Doesn't forwarding every email get tedious?

Less than you'd think. Most people forward two or three emails a day — the ones with real commitments in them — and BCC the capture address when making a promise, which takes zero extra clicks once it's in your autocomplete. For everything else, there's Cmd+K natural-language quick add. And for people who want true autopilot, opt-in ambient inbox sync is planned on the roadmap.

What about commitments made in Slack or meetings?

Today, paste or quick-add them — the same natural-language parser handles “send Dave the deck by Friday” from any source. Slack/Teams capture and meeting-transcript extraction are both on the roadmap; email came first because that's where the costliest drops happen.

Can the AI add tasks to my list on its own?

No, and that's a design principle, not a limitation. Every extraction is a candidate until you accept it. Your list never grows without your yes — which also means a malicious email can't plant tasks in your day.

What happens to the emails I forward?

Processed, then purged. We keep the extracted commitments and a short snippet for context; the raw email is deleted, and nothing is ever used to train AI models. The details are on the security page.

Next up: the Follow-Up Radar chases what you captured, and the command center tells you which loop to close first. Or see pricing — free during early access.

Your inbox already wrote the to-do list.

Forward one email and watch it become a tracked commitment. Free during early access — no credit card, no inbox access.