Getting started

Arcus connects to a data warehouse you already have, and lets the people who own a P&L ask questions in plain English. You get back a narrative, a chart, and the underlying rows — every number cited.

This guide walks you from a fresh signup to your first answer. Allow about 20 minutes if you have warehouse credentials handy.

Before you start

You need three things:

• A warehouse Arcus can read: BigQuery, Snowflake, or Postgres.
• Read-only credentials for the schema you want to expose. We never write to your warehouse, but we expect you to scope us tightly anyway.
• About 20 minutes of your time, plus 10 minutes from whoever owns warehouse access if that isn't you.

You do not need to clean your data, build a model, or pre-define metrics. Arcus discovers the schema and proposes a semantic layer for you to review.

Step 1 — Sign up

Go to usearcus.ai/signup and create an account with your work email. We'll send a confirmation link. Click it, set a password, and you'll land in your new workspace.

Your first action in the workspace is to name your organization. This is what your team will see in the workspace switcher. Most teams use their company name.

Step 2 — Connect a warehouse

From the left sidebar, open Settings → Connections and click Add a connection. Pick your warehouse type. The forms ask only for what's needed:

• BigQuery — paste a service account JSON key with BigQuery Data Viewer and BigQuery Job User roles. See Connect BigQuery for the GCP-side setup.
• Snowflake — provide an account identifier, warehouse name, and key-pair credentials for a read-only role. See Connect Snowflake.
• Postgres — provide a connection string for a read-only user. See Connect Postgres.

When you save, Arcus runs a SELECT 1 (or equivalent) to verify the credentials. You'll see a green Connected badge if it worked. If not, the error tells you exactly what to fix.

Step 3 — Discover the schema

Click Discover schema on the connection. Arcus reads your information schema and lists every dataset, table, and column. For tables with a date column, we also surface the range so you can see how much history you have.

Pick the tables you want exposed to questions. Most teams start with their analytics-ready mart.* or dbt_* schemas and skip raw event tables. You can always add more later.

Step 4 — Approve the semantic layer

This is the step that makes the rest of the product feel intelligent. Arcus reads the discovered schema, samples some rows, and proposes a semantic layer — a YAML file describing your entities, metrics, dimensions, and synonyms.

For example, on a marketing warehouse Arcus might propose:

entities:
  order:
    table: mart.orders
    metrics:
      revenue: SUM(gross_revenue)
      orders: COUNT(*)
    dimensions:
      country: country_code
      channel: sales_channel
    synonyms:
      revenue: [sales, gmv]
      country: [geo, region]

Read it. Edit anything that's wrong. The proposal is a starting point, not a final answer — you know your data better than we do. Once it looks right, click Approve. The layer is versioned per-tenant and can be re-edited at any time.

For more on what goes in here, see Define a metric.

Step 5 — Ask your first question

Click Chat in the sidebar. Type a question the way you'd ask a smart analyst:

What was revenue last week, broken out by country?

Hit enter. Arcus shows three things:

1. A narrative. Two to four sentences in plain English with a confidence tag (high, medium, or low).
2. A chart. Auto-selected based on the shape of the answer — line, bar, stacked, or table.
3. The rows. Click "show data" to see the underlying result set, with the SQL Arcus generated and a link to the audit trail.

Every number in the narrative cites the column and filter that produced it. Hover over any figure to see the source.

For tips on how to phrase questions that get good answers, see Ask your first question.

Step 6 — Invite your team

From Settings → Team, invite the people who'll use Arcus. Two roles in v1:

• Admin — full access. Can manage connections, edit the semantic layer, change billing.
• Member — chat and dashboards only. No settings access.

Invitees get an email with a magic link to join.

What's next

• Understanding the answer card — the three-layer format, in detail.
• Pin to Watch — save important answers and get them refreshed automatically.
• Automations — turn a question into a recurring brief or alert.

If anything in onboarding is unclear, email hello@usearcus.ai — a real person will reply.