This walks a new member from nothing to a working query and a connected agent. Each step names the button you’re looking for.
On a brand-new tenant, the Overview page covers this same ground itself. Its Get started card leads with one prompt for your coding agent (see Your first run); Set up by hand instead opens the step-by-step checklist: Connect your data, Bring your terminal and AI tool, and Build your artifact, publish it as an app. Each step marks itself done as your tenant’s own data shows it happened, not by you checking a box.
1

Sign in

Go to your tenant’s URL (for example acme.app.us.dippr.ai) and sign in with your password, or with “Sign in with SSO” if your tenant uses one. If you got here from an invitation link instead, open it, choose a password, and click Go to sign in. Don’t have a tenant to join? See Create a tenant to sign up on your own, with no invitation needed.
The sign-in page

The sign-in page.

2

Be added, or accept an invite

Someone with the account_admin role adds you from Settings → Users, clicking Invite user and choosing your roles. Dippr doesn’t send an email: the admin copies the one-time link and sends it to you themselves. The link expires if unused; opening it and setting a password is what creates your account.
3

Connect a source

Go to Data sources and open the Add integration tab. Click the card for the kind you want (Postgres, Snowflake, ClickHouse, Google Sheets, DuckDB, or an Iceberg REST catalog); its form opens already set to that kind. Fill in the connection details and click Connect. When the source can be probed, Connect tests it first and won’t save if that test fails, unless you explicitly check Register anyway, without a verified connection — so a saved integration is normally one Dippr actually reached, not just one you typed.The Capabilities column on the Connected row is set automatically from the source kind — you don’t choose them. Every kind that can be queried is marked Queryable from engines by default, so a database source becomes a catalog on your engines without any extra step. (A cloud-credential kind, which nothing queries directly, simply doesn’t offer that capability.)Note which data plane you register the source on. An engine only sees sources on its own plane, so if your tenant has more than one, pick the same plane here that you’ll create the engine on in the next steps — otherwise the engine’s catalog set comes up empty.
No source to connect yet? On the Overview page’s Get started card, click Use sample data. Dippr adds a small read-only avocado prices sample dataset as a catalog, hosted for you — no credentials, no file to upload — so you can reach a first query immediately. Once you have an engine, query it with SELECT * FROM <catalog>.main.avocado_prices LIMIT 5 (the catalog is named for the sample). Remove it any time from the Data sources page.
4

Confirm your data plane is connected

Most tenants start on Dippr’s own data plane, already connected. Check Settings → Data planes for a Connected status badge. If your tenant runs a separated data plane inside your own cloud account, this is where you confirm it has come online before continuing.
Settings › Data planes: the registered planes and their connection status

Settings › Data planes: the registered planes and their connection status.

5

Create an engine

Go to Settings → Query engines and click Create engine. Choose a size from the ones your plan allows (xs on Free; up to m on Team; up to xl or custom on Enterprise; see Entitlements) and the data plane to run it in — the same plane as the source you just connected, since an engine only serves sources on its own plane. The engine’s first catalog set, built from the queryable sources on that plane, applies automatically. Wait for its power state to reach Ready before you query it.
6

Apply catalogs

Connecting a queryable source, or a new workspace database starting to serve, changes what an engine’s catalogs would be, but an engine only reads catalogs at startup. When a row shows pending changes, click Apply, then confirm Apply and restart. This briefly restarts the engine and interrupts anything running on it, so do it when nothing time-sensitive is mid-query.
7

Run your first query

Open Explore, open a new query tab, write a statement against a catalog you connected, and press Run (⌘↵). Every queryable source you connected and applied on this engine’s plane appears as a catalog here. Results stream to your browser straight from your data plane. From here you can save the query, or open one of the tables you queried and switch to its Lineage tab to see what feeds it.
8

Connect Claude via MCP

From a terminal, install dip and sign in:
Signing in registers your tenant’s MCP servers with the coding agents on your machine as its last step, so Claude (or Codex, or Cursor) can immediately run SQL and search your metadata as you. Run it again any time with dip mcp install. See Install dip and sign in.
9

Optional: share a data app

Ask your connected agent to build a Dippr app your team can open. Preview it, then publish with dip app push and run dip app install. Confirm the exact version’s access and code trust once; dip waits for a ready app and gives you its link. Members can publish and install their own static apps by default. Each colleague opening the link queries with their own access. Docker and buildx are required on the publishing computer.No terminal? In Claude or ChatGPT on the web, ask your AI: “Publish this as a Dippr app.” It publishes the page as you through the Dippr connector and gives you the link. See Publishing an app from chat.See Build and share your first app for the whole journey, step by step, and Install an app for the console alternative.

Where next

How Dippr fits together

The full vocabulary behind every step above.

Roles and personas

Who on your team needs which of these sections.

Semantic layer

Turn your tables into certified metrics everyone can trust.

MCP and AI agents

What an agent can and can’t do once it’s connected.