Looky Docs

What you can achieve after this docs path

If you are new to Looky, this docs set takes you from zero to a published workspace with at least one dashboard visible in https://my.looky.studio. It is written for builders, analytics engineers, and delivery leads who need repeatable execution, not theory.

Build focused answers, not giant dashboards

Most BI tools push you toward one dashboard that tries to answer everything. The result is something nobody reads. Looky is built around the opposite: one focused dashboard per business question.

A dashboard that answers "Where is fulfillment leaking this month?" is something a team will open every morning. A dashboard with 40 charts and no clear question is a loading screen.

This shapes everything in the build workflow:

• Models define your semantic layer once. One measure definition, reused across every chart that needs it.
• Visualizations bind one query to one chart. No logic, no duplication.
• Dashboards compose visualizations around a single question — small enough that the answer is obvious, specific enough to be actionable.
• Document mode takes that focused dashboard and makes it a deliverable: a top-to-bottom report that exports to PDF on a schedule and lands in the right hands without anyone having to open the UI.
• Cross-filtering in interactive dashboards lets users click any bar or point to slice all charts simultaneously — the kind of exploration that surfaces answers that no single chart could show alone.

A well-built workspace makes this concrete: several separate dashboards, each answering one distinct question about the business, rather than a single sprawling surface trying to cover everything.

Zero-to-dashboard path (45 to 90 minutes)

1. Accept invitation and verify role scope in the UI (Owner Invitation).
2. Sign in to https://my.looky.studio and link CLI to your local root (Sign In and Connect the CLI).
3. Set up BigQuery service account and wire credentials into workspace runtime config (BigQuery Dataset Access).
4. Set billing context and create your first workspace (Create Your First Workspace).
5. Author sources, models, visualizations, and dashboards (Build Workflow).
6. Run looky validate, looky diff, looky push and verify in UI.

Command sequence most teams run daily

looky login https://my.looky.studio <local_root>
looky whoami

cd <local_root>
looky billing list
looky billing use <billing_account_id>

cd <local_root>/<billing_account_id>
looky create <workspace_slug> --name "My Workspace"

cd <workspace_slug>
looky status
looky validate
looky diff
looky push
looky list visualizations
looky list dashboards

What makes this docs set different

• Every phase includes success criteria, not only command examples.
• The workflow is end-to-end: access, BigQuery setup, authoring, publish, verify.
• Examples are grounded in real workspace structure and real CLI commands you will actually run.
• Troubleshooting is organized by symptom with deterministic fix steps.

Reading these docs with an LLM or coding assistant

The same content you read in this site is also published in agent-friendly formats, served from the same source — there is no separate copy to drift. Two URL conventions:

• Per-page Markdown. Append .md to any docs URL to get a clean Markdown rendering with YAML frontmatter (title, slug, language, summary, last-modified date).

  https://looky.studio/en/docs/build-workflow/viz-types/bar.md
  https://looky.studio/en/docs/build-workflow/cross-filtering.md
  https://looky.studio/en/docs/reference/source-adapters.md

• Indexes following the llmstxt.org convention.
  • https://looky.studio/en/llms.txt — short index of every docs page in this language, grouped by section, with a one-line summary per page and a link to its .md version. Spanish at /es/llms.txt.
  • https://looky.studio/en/llms-full.txt — concatenation of every docs page rendered as Markdown, separated by ---PAGE--- boundaries. One fetch and the assistant has the entire docs set.
  • https://looky.studio/llms.txt — root index, redirects to the default language.

Use the per-page .md when you (or your assistant) only need the reference for one feature; use llms-full.txt when you want the assistant to have the whole context up front.

The format is stable: page slugs do not change once published, so any URL captured today keeps working as the docs grow. Authored content is HTML; the Markdown and indexes are derived at request time from the same MongoDB store, so they cannot drift from what humans see in the browser.