Getting started

Luma BI is a managed platform — we run a dedicated Postgres + Redis + Express stack on isolated infrastructure for you. There's no install step on your side. This guide describes what happens between kickoff and your first live dashboards (typically a week for Starter, 1–2 weeks for Growth).

For technical depth on how the platform is built, see Deployment and the Schema reference.

What we need from you on day one

A point-of-contact with admin authority on at least one of: your data warehouse, your file exports, or your billing system.
Your SSO domain(s) — e.g. acme.com. We restrict sign-in to email addresses on these domains.
Your industry template — marketplace, e-commerce, SaaS, fintech, travel, subscription, healthcare, or “custom”.
Either a warehouse credential or a sample data export. See “Data sources” below.

Data sources

Pick the path of least resistance. You can mix and match later.

Path A — Warehouse connector (recommended for Growth+)

We register a read-only role on your warehouse and pull data on a schedule into your dedicated Luma instance. Once synced, dashboard queries hit our managed Postgres — never your warehouse — so you don't pay per query and dashboards stay sub-second.

| Warehouse | Auth | Notes | | --- | --- | --- | | BigQuery | Service account JSON | Read-only role on your project / dataset | | Redshift | IAM credentials or DB user | Either VPC-peered or via public endpoint with allowlist | | Snowflake | OAuth or username + key-pair | Read-only role + warehouse | | Postgres | DB user | TLS required | | MySQL | DB user | TLS required |

Path B — File drops

If you don't have a warehouse, or just want to start fast: upload CSV or XLSX exports through the admin UI. Filenames don't matter — the importer detects file type from headers. SHA-256 dedup means re-uploads are no-ops.

Path C — SaaS API connectors (Growth+)

For event / transaction data that lives in SaaS tools rather than a warehouse: Stripe, GA4, HubSpot, Segment, Shopify, Mixpanel. Same model — we sync into your dedicated Postgres so dashboards don't consume your API quota.

What you'll see when it's live

Your admin sign-in URL (something like acme.luma-bi.com)
Google SSO restricted to your configured domain(s)
25+ dashboards rendered against your data, sliceable by country + date range
The AI Analyst page — admin-gated for first creation, all-user-readable for results

First-week checks

Executive dashboard should show non-zero numbers — if it doesn't, your sync probably failed or your filter is too narrow.
AI Analyst — try an example like “Which channels convert best to revenue in the last 30 days?”
Admin · Users (admin-only) tracks who's logged in and what they've viewed.
Admin · AI Spend (admin-only) shows the running Anthropic bill so you can see the cost guardrails working.

Common questions

Question	Answer
Can we self-host?	No — Luma BI is a managed offering. We run a dedicated, isolated stack for each customer.
Where is our data hosted?	UK by default. EU and US regions available on Growth and Enterprise.
Will dashboards hit our warehouse repeatedly?	No. Data is synced into your dedicated Luma Postgres on a schedule. Dashboards query that, not your warehouse.
Can the AI Analyst write data?	No. The agent only has a read-only SQL tool — DDL / DML are rejected by the validator.
What happens if we hit the AI cost cap?	New AI Analyst runs are paused for the day. Errored runs don't count against the quota.

Ready for more depth? See the AI Analyst architecture or dive into the schema.