Read time: ~5 minutes · Series: Platform Admin getting started · Next: Analyst vs platform divide
Not a warehouse, not an ETL scheduler
Some assume Lantide means "load all data in" or "configure nightly jobs." It is an analyst-local query and collaboration IDE:
- Data stays in your Postgres, shared CSV folders, etc.
- DuckDB on the user's machine queries and JOINs
- Analysis artifacts (Plan, Report, project knowledge) live in app data dirs, portable via
.lantide
Not responsible for: nightly ETL, replacing BI permission models, multi-user shared runtime.
Three data layers (ops view)
| Layer | Holds | Your typical ops |
|---|---|---|
| Local data files | CSV, Excel, Parquet, etc. | Workspace data folder points at team folder |
| External connections | PostgreSQL / MySQL / SQLite | Connections attach; query as alias.schema.table |
| Application state | Workspace settings, SQL tabs, projects, connection profiles | Workspace / Project .lantide export/import |
Cached tables are local materializations after queries—part of workspace state; after migration often need Source Run rebuild.
vs BI and generic AI
- BI: monitors defined metrics; Lantide for adhoc sign-off review (Business & Ops: vs BI)
- ChatGPT etc.: no workspace, no Plan state; Lantide is an environment-backed analysis Harness
Product positioning: Series guide §1.
One thing to do now
Sketch for the team: company Postgres / folders → Lantide workspace (connections + load dir) → analyst projects. Mark what you maintain vs what analysts own.