Quickstart

From download to your first ranked list of savings. Budget about fifteen minutes, most of which is the installer working while you watch a log.

Before you start, check you have what Requirements lists. The short version: an account admin login, a Unity Catalog workspace with system.* enabled, a SQL warehouse, and the Databricks CLI on your machine.

Register at lakemine.ai

Create your account at lakemine.ai , or create it from the installer’s sign-in screen, which registers you through the same portal and starts a 14-day trial.

There is no account id or license key to keep track of. Your identity travels with the session you sign in with.

Download and launch the installer

Get the desktop installer for your platform and open it. On macOS the first open needs one extra step, covered in Download and install.

Email and password, on the installer’s first screen. This binds the install to your billing account and confirms your subscription is active.

Installer, sign-in panel. Empty email and password fields, “Sign in” button, and the “Don’t have an account? Create one” switch beneath it.

Link your Databricks account

Pick your cloud, enter your Databricks account id, and sign in through the browser as an account admin. The installer confirms your admin rights and lists your workspaces.

Your credentials stay on your machine. Lakemine never receives them.

Installer, “Link your Databricks account” panel. Cloud dropdown set to AWS, account id filled in, and the account checks list showing “Account admin (can list workspaces)” passing with a green tick.

Choose the hub workspace

The appliance deploys into one workspace and analyzes the whole account from there. Pick one that already has a SQL warehouse and Unity Catalog attached.

If any workspaces sit outside the hub’s region, the installer tells you now, before you deploy. Those get cost-only coverage.

Then sign in to that workspace. This is a second browser sign-in, at the workspace level.

Installer, “Choose the hub workspace” panel. Workspace dropdown open showing several workspaces, region hint below it, and the region-coverage warning visible for an account with an off-region workspace.

Choose the deployment destination

Pick where the results go and how often the scan runs:

Catalog and schema, default schema name lakemine
SQL warehouse, or let the installer create a small serverless lakemine-wh
Scan window, default the last 30 days
Scan schedule, default daily at 06:00 UTC
Dashboard viewers group, the group whose members can open the cost dashboard

Installer, “Deployment destination” panel, fully filled in. Catalog and schema set, scan window on “Last 30 days”, schedule on “Daily at 06:00 UTC”, a warehouse selected, and the viewers group dropdown showing a chosen group.

Let it verify

The installer runs a pre-flight so the deploy cannot fail half-way. It checks the Databricks CLI locally, then workspace authentication, the catalog, the SQL warehouse, service principal listability, and whether the system.* schemas are enabled.

Blocking checks must pass before Deploy turns on. Missing system.* schemas are a warning rather than a blocker, and the installer names which detections will be skipped.

Installer, “Workspace verification” panel with every check passed, including the system.* check, and the Deploy button enabled.

Deploy

The last panel starts on entry and streams every step into one log: rendering the dashboard, preparing the lakemine-scan service principal, building and uploading the compiled wheel, deploying the job and dashboard, applying the grants, and starting the first scan.

Each grant’s outcome appears in the same log as verified, applied, or needing a manual step with the exact command printed for you to hand to a metastore admin.

Installer, install panel, log mid-run with several completed steps visible and one in progress, showing at least one grant line and its outcome.

Read your first findings

The first scan starts as part of the install. Once it finishes, open “Databricks Cost Optimization” in your workspace’s dashboards.

Start on Account Overview: total spend analyzed, estimated savings, and the top opportunities ranked by dollar impact. Then use Workspace Detail to drill into any one workspace.

Databricks Lakeview dashboard, “Databricks Cost Optimization”, Account Overview page, populated with real scan data. Rollup tiles, cost and savings by workspace, and the ranked top-opportunities table.

If the scheduled scan has not run yet, trigger the lakemine-scan job once by hand to get data immediately. After that it runs on the schedule you chose.

What you should see

A ranked list of findings, each with a dollar figure attached, sorted so the largest recoverable spend is at the top. Findings carry a severity so you can tell a high item that is costing you real money from a low hygiene issue.

The dollars are estimates derived from your actual billed usage over the scan window, not list-price arithmetic. The Engines section shows the calculation behind each one.

If something is missing

No findings at all. Check the lakemine-scan job ran successfully. If it reports “no subscription key found”, the lakemine secret scope or its key was removed. Re-run the installer, or use Reconnect to rewrite it without a full redeploy.

Some detections skipped. A system.* schema is not granted. The job names which ones it skipped, and the fix is a metastore admin applying the SELECT grant.

Workspace ids instead of names. system.access is not granted. This is an automatic fallback and nothing else is affected.

Cost-only coverage on some workspaces. Those workspaces are outside your hub’s region. system.billing is global so spend is covered everywhere, but the compute, query, and lakeflow telemetry is regional. A per-region hub extends coverage.

Installation covers every step above in detail.
The Connecting Databricks section covers grants, identities, and revocation.