Getting Started: Setup + Your First Goal

Most people set up Muse the way they set up any app — tap through, connect everything, ask it something vague — and then wonder why it feels underwhelming. The first hour matters more than it looks. Here's the version that actually works.

Step 1: Install (10 minutes)

Pick your surface: the iOS or Android app, muse.ai on the web, or WhatsApp (message Muse like a contact). US only, 18 and older, as of this writing.

Sign-up asks for a payment card even on the free tier — it's there so rising usage doesn't interrupt your work, not a charge. You'll also meet the usage meter, which shows your remaining allowance as a percentage. Glance at it now so you recognize it later.

Step 2: Make it yours (5 minutes)

Before any real work, do the small personalization pass: give Muse a name, pick an avatar, and set the proactivity level. This isn't vanity — naming it changes how you brief it (you write to someone, not something), and proactivity controls whether it surfaces suggestions unprompted or stays quiet until asked. Start conservative; you can turn it up later.

Step 3: Connect exactly one service (10 minutes)

Ask Muse ("connect my Gmail") or go to Settings → Connectors, tap Connect, read what's shown, and authorize. Just one. Email is the best first connector for most people — it's where the highest-leverage early tasks live.

Read the access description before you tap through. This is also the moment to internalize the rule: a connector is a door you opened. You'll add more doors later, deliberately.

Step 4: Run your first delegation (20 minutes)

Pick one job you already understand — today's unread email is the canonical choice. Write the brief in three parts:

1. Outcome: "Go through today's unread email. Give me three buckets: needs my reply today, can wait, junk. For each item in bucket one, one line on why."
2. Boundaries: "Don't send, delete, or archive anything. Don't open attachments from unknown senders."
3. Check-in: "Show me the summary and stop. I'll tell you what to do next."

Read the result like an editor, not a fan. Where was it wrong? The most valuable sentence you'll write this week is the correction: "treat anything from someone I've emailed before as personal, regardless of formatting" fixed my triage permanently.

Step 5: Set your first goal (10 minutes)

Now the part chatbots can't do. Give Muse something ongoing: "Every weekday at 8am, summarize my unread email from the last 24 hours the same way we just did." Check the Goals tab to see it tracked, and Upcoming for the schedule.

Then — and this is the discipline — watch the first two runs before you trust it. A recurring task with a bad brief fails on a schedule. Two reviewed runs are the price of admission for automation.

Step 6: Learn the approval ladder (ongoing)

The first time Muse asks to do something sensitive, you'll see the approval options: allow once, for the task, for the site, always for the connector, or deny. Default to allow once. Widen only for patterns you've watched work. Peek at the Activity log (tap your assistant avatar) after the first few tasks — it's the fastest way to understand what the agent actually does when you're not watching.

Mistakes to skip

• Connecting everything on day one. Each connector is surface area. Earn them one at a time.
• Vague briefs. "Help with email" gets you a lecture about email. Outcomes, boundaries, check-ins.
• Automating before reviewing. Two clean runs before anything recurring. No exceptions.
• Reflex approvals. Read the prompt — especially amounts, recipients, and unfamiliar sites.
• Pre-paying. Run free until the usage meter tells you otherwise. The upgrade takes seconds.

Do the hour above and you'll be past the stage where most people stall — an installed app they ask trivia. You'll have a working delegation, a goal running, and a feel for the approval rhythm. That's the whole game; everything else is repetition at larger scale.

Quick answers