Skip to content

PRIVACY.md

Latest commit

 

History

History
81 lines (57 loc) · 3.67 KB

PRIVACY.md

File metadata and controls

81 lines (57 loc) · 3.67 KB
Raw

Privacy & Telemetry

Databricks Solution Builder ships with opt-out anonymous usage analytics. This page explains exactly what's collected, why, and how to turn it off.

TL;DR

  • Analytics are on by default when you run the app.
  • Only aggregated, anonymized events are sent — never PII, secrets, source code, prompts, or your data.
  • We use the data only for product improvement — never for sales contact, marketing, or any commercial purpose.
  • The underlying dbdemos-tracker package filters at the source: only @databricks.com users send events. External installations send nothing in practice, but the middleware is still registered.
  • To disable entirely, set DEMO_PROMPT_GENERATOR_TRACKER_ENABLED=0 in your .env.

What's collected

The dbdemos-tracker middleware fires on app-view events (page loads, navigation). Each event carries:

Field Example Purpose
Project identifier industry-demo-prompts Distinguish events from this app vs. other dbdemos installations
App version 0.1.10 Correlate behaviour to release
Page / route hit /projects/<uuid> Understand which features get used
Anonymized user hash (hashed) Distinct-user counting; not reversible to an email
Coarse timestamp 2026-05-12 Trend analysis only

The tracker does not collect:

  • ❌ Source code, file contents, project files, or chat transcripts
  • ❌ Customer data, workspace data, or any data from your Databricks workspace
  • ❌ Anthropic prompts, completions, or agent traces
  • ❌ Authentication tokens, OAuth credentials, or environment variables
  • ❌ IP addresses (filtered at the receiver)
  • ❌ Free-text user input of any kind

Why we collect it

The single goal is product improvement: knowing which features are used (and which aren't) so we focus engineering effort where it matters. Specifically:

  • Detect dead features so we can remove them.
  • Detect popular features so we can invest in them.
  • Catch regressions when usage of a path drops after a release.

We never use this data for sales contact, lead scoring, marketing, or any commercial purpose. Sales never see it.

How to opt out

Local development

Set the env var in app/.env:

DEMO_PROMPT_GENERATOR_TRACKER_ENABLED=0

Databricks App deployment

Add the same env var to databricks.prod.yml's env: block under the relevant target:

env:
  DEMO_PROMPT_GENERATOR_TRACKER_ENABLED: "0"

Then redeploy: databricks bundle deploy.

Electron desktop app

Set the env var before launching the packaged binary:

DEMO_PROMPT_GENERATOR_TRACKER_ENABLED=0 open /Applications/DatabricksSolutionBuilder.app

In all cases the middleware is short-circuited at boot in app/src/demo_prompt_generator/backend/core/_factory.py — no network call is ever made.

Auditing the code

The entire tracker integration lives in two places:

  1. app/pyproject.toml — the dbdemos-tracker dependency line.
  2. app/src/demo_prompt_generator/backend/core/_factory.py (lines 82–106) — the conditional registration.

The dbdemos-tracker package itself is open-source on PyPI.

Changes to this policy

Any change that expands what's collected will be called out in CHANGELOG.md (when we add one) and in the release notes of the affected version. Reductions in scope just happen.

If you spot something in the code that contradicts this document, please open an issue — we'll fix the code, the doc, or both.