RaziConsulting Services
Menu
DataPulse AI · a Databricks App

The Databricks platform team in a box.

Spend, Query Tuning, Data Cleansing and Pipelines in one app that runs inside your own Databricks workspace. It finds waste and failures, then fixes them with changes you preview, approve, verify and, where possible, undo.

Book a demoHow it stays in your account
The DataPulse AI Spend page: spend, forecast, verified savings, spot ratio and idle waste cards above a spend trend chart and chargeback by tag.
Spend page · sample data, not a customer's bill

Runs in your Databricks account

  • Zero egress. No external service, no API key and no telemetry. The only model it calls is a Model Serving endpoint in your workspace.
  • Reads as the signed-in user. Unity Catalog permissions, row filters and column masks apply.
  • Its own identity has no grant on your catalogs or tables. The app's service principal writes the app's state and runs approved actions.
  • Every change is an action. Preview, approval by role and risk, execute, verify and roll back, recorded in an append-only ledger.
  • No personal access tokens. The user's token is never logged or stored.
  • Licence checked offline. No phone-home and no usage metering.

Read the security overview

One app

Three modules for the platform team.

Each module builds on what Databricks already records (system tables, Unity Catalog and lineage) and proposes changes you approve. An Overview page ties them together: verified savings, open incidents and time to resolve, the data-quality trend and pending approvals, each number with a link to how it is calculated.

01

Spend + Query Tuning

Where the money went and where it is heading: cost by SKU, chargeback by tag, anomalies and a month-end forecast. Budget guardrails propose a fix for a person to approve, and the slowest queries get an AI rewrite to test.

Spend and Query Tuning
02

Data Cleansing

Turn a raw bronze table into a clean silver table. Describe the fix in plain English, review the SQL behind a safety gate, preview 100 rows, then write a new table. Your source table is never changed.

Data Cleansing
03

Pipelines

One board for every Lakeflow job and declarative pipeline across your workspaces: failing and stalled work, the cost of failed and retried runs, hygiene checks, AI root cause and repair actions.

Pipelines
Pipeline Health Center

Every failed run has a cost. See it, then fix it.

The Pipelines board compares each job and declarative pipeline with its own history, not a fixed threshold, across all your workspaces.

  • The dollars and DBUs spent on failed and retried runs, per job and in total.
  • Stalled work flagged when a run takes more than twice its own p90.
  • Hygiene checks: no timeout, no retries, no owner, scheduled work on all-purpose compute.
  • AI root cause from the run's error text. Your table rows are not sent to the model.
  • Repair Run, Retry update and Full refresh of one table, each through approval.
  • Downstream impact from Unity Catalog lineage before you act.
  • On the Business tier: data SLAs, and incidents with owners and time to resolve.

Explore Pipelines

The DataPulse AI Pipelines board: workloads, failing, stalled, SLA breaches, open incidents and wasted cost cards above a job-by-day health matrix.
Pipelines board · sample data
Acts as well as reports

Every change follows one path.

Whether it starts on Spend, Data Cleansing or Pipelines, a change is an action with a preview, an approval and a ledger row. A saving counts only after it has been measured.

  1. PreviewThe change, its risk, the downstream impact and how to roll it back.
  2. ApproveBy role and risk: an operator (or a cleansing engineer, for Data Cleansing) for low risk, an admin for high risk.
  3. ExecuteRecorded in the ledger first. If the ledger cannot be written, nothing runs.
  4. VerifyAfter the observation window, a person presses Verify and the effect is measured.
  5. Roll backRestores the previous setting. Retry update and Full refresh cannot be undone, and the dialog says so.
Approval is the default. Only a budget an admin sets to Auto-Idle (Business tier) lets low-risk actions on idle resources run without a person, and each one is still recorded. Guardrails never cap your bill.
Who it is for

Built for the people who run Databricks.

Platform leads

The platform team in a box, inside your workspace. No vendor SaaS reads your billing or metadata.

FinOps

Chargeback by tag, budgets that propose a fix, and a ledger where only measured actions count as savings.

Data engineering

Failures with their cost and cause, repairs you approve, and a clean silver table without writing the SQL.

Data leaders

A verified return on one page, without another vendor holding your billing data.

See DataPulse AI on your own workspace.

A 30-minute walk-through of Spend, Data Cleansing and Pipelines, then a read-only Free Health Check on one of your workspaces if it fits.

Book a demoSee pricing