What Tend actually does.

Six things carry the product. Everything else exists to make those six work.

Now Live today
Beta Being built — not yet switchable
Planned On the roadmap, no date
The core · Now

The goal audit

Give Tend a goal. It reads your real history and hands back the ladder — who has leverage on this goal, the next move for each of them, and who to stop chasing. Every line carries the conversation it came from.

  • Ranked by real leverage, not by who you spoke to last
  • Says who not to spend attention on
  • Renders as it writes — summary in seconds, then each person
  • Re-runs cheaply; it remembers who it already judged
Trust · Now

Every claim shows its receipt

Tend never asserts anything it can't point to. Every audit line, every suggested move, every “you owe / you're owed” traces back to the exact conversation it came from — so you're never taking a black box's word for it.

  • If Tend can't show the receipt, it doesn't say it
  • Grounded in your own history — never the public internet, never bought data
  • A source it invents is a source that doesn't render — hallucinated citations are dropped, not shown
Pipeline · Now

Stages you name yourself

A raise and a hire don't run the same pipeline, so Tend doesn't pretend they do. You name the stages; each keeps a type underneath, so your labels never break the progress maths.

  • A board per goal, not one pipeline for everything
  • Tend proposes a stage set; nothing applies until you save
  • Moves proposed from real replies — a decline goes to Dead, not forward
Capture · Now

The history builds itself

You never hand-log a conversation. Email, calendar and LinkedIn flow in on their own, which is what makes every AI claim traceable to something you actually sent or received.

  • Gmail, Outlook, or any IMAP mailbox including iCloud
  • LinkedIn threads via the Chrome extension — view-triggered, never background-scraped
  • Your full LinkedIn export in one go; strangers land dormant
Every morning · Now

A brief that leads with the one thing that matters

Where your goals stand and today's few moves, opening with the single most consequential one. Every item links to a real person and a real goal.

  • Free, and never spends credits — you shouldn't be stopped by something you didn't choose to run
  • Drafts attached, ready to send
Ask · Now

Ask your network a plain question

Questions about the people you actually know, answered from your real history — every name linked to their record, with the reason attached.

  • Never the public internet, never a bought database
  • If your network can't answer, it says so instead of inventing someone
Meetings · Now

The meeting loop — prepped before, closed after

Your brief carries today's meetings: what you owe each person, where they sit on your goals, and the ask — with a briefing even when an attendee isn't a contact yet. Then the half nobody else ships: the morning after, Tend asks how it went, and the answer lands on the record.

  • A calendar page of its own — agenda, week and month views
  • “This meeting ladders to nothing” — the flag no cadence tool can raise
  • The close: your one-line answer becomes commitments and next moves, receipts attached
  • Meeting notes stay editable — the close is a surface, not a one-shot form
Agent access · Now

Tend, wherever your assistant is

Connect Claude — desktop, web, or the phone in your pocket — and it answers from your Tend record. One URL, click Allow, done. No token, no config file. And it never spends your credits: the thinking runs on your assistant's plan, not your allowance.

  • Paste a job spec or an email and ask who in your network is relevant — the one thing Tend's own Ask can't do, because it never sees your clipboard
  • Read-only is a hard boundary — it can read goals, commitments and history; it cannot create, edit or send
  • Every connected device listed in Settings, each with its own Revoke
  • Speaks MCP, so any client that does works. Claude is first-class; ChatGPT needs Developer mode
Also in the box

The things that make those eight usable

Less headline, no less load-bearing.

Drafts in your voice Now

Channel-aware for email, text or LinkedIn — grounded in the real thread and matched to how you actually write.

Commitment tracking Now

What you promised and what you're owed, pulled out of real conversations with the receipt attached.

Fit scores Now

Suggested people carry a score out of 100 — role fit and access, shown separately. It's allowed to come back low.

Honest gap diagnosis Now

The audit names who's missing, checks your dormant imports for a match, and hands you a plain-keyword LinkedIn search plus a warm path through someone you already know — never a purchased list.

The weekly review Now

Every goal re-ranked overnight, on a schedule — and a warning before a manual re-run buys a second copy of what the week already paid for.

Relationship synthesis Now

Warmth, recurring topics and open threads for any contact — a briefing before you walk into the room.

Todoist sync Now

Connect your own Todoist and Tend's reminders land in the task list you already look at, with a clear reconnect path.

People in your history Now

Finds people who were named but never added — the offered introduction, the person who signs off — each with a receipt.

Mailbox scan Now

Surfaces who you actually correspond with from envelope data alone, with bulk promote — so you don't start from an empty app.

One-click send Beta

Send an approved draft straight from Tend through Gmail or Outlook, instead of copying it into your mail client.

The brief on a schedule Now

On the paid plan the brief runs on a schedule and is waiting when you arrive — and it never spends credits, whichever way it runs.

Mobile app Planned

A native app, eventually — no date. Meanwhile Tend already answers on your phone through Claude: connect it under Agent access and ask from anywhere. And Tend runs in any mobile browser today.

And the rest

The everyday things that have to be right

Unglamorous, and the product doesn't work without them.

Contacts, groups, tags and custom fields
Important dates and reminders
Cadences, for the few worth a rhythm
Activity timeline per person
⌘K search, reaching dormant imports
Duplicate detection and reversible merge
Guided goal intake — “sharpen this goal”
Pace, as a goal-momentum metric
Enrichment you approve, never bought data
“I sent this”, for channels Tend can't see
Import from CSV, vCard, Google or Outlook
Export everything to CSV or JSON
One-click account deletion that deletes
Keep any thread or contact out of AI entirely
An itemised credit statement, in credits
Encrypted tokens, per-account isolation
Light and dark, following your system
Help chat that knows the product, free
Chrome extension from the Web Store Beta
Copied-in (cc) messages captured Beta

On beta and planned. Beta means specified and being built — shown here so you can see where Tend is heading, not something you can switch on today. Planned means it's on the roadmap with no date. We'd rather show you the direction than pretend the product is finished.

Start with what's live today.

Free to start, no card, and about 4 goal audits included so you can run one. The $49 plan makes them recurring — 200 credits a month, or $39/month billed annually — and adds the daily brief.

Get started