Read time: ~8 minutes · Role: Platform Admin (Data Engineering) · Next: Work through the reading path below article by article, or walk through one test connection and Workspace export.
Questions you may be asking
"The analyst wants to connect the company PostgreSQL and also drag in Excel—isn't this tool local?"
"How do we move the workspace when switching computers? Can we just copy folders manually?"
"Do I need to modify the Agent code?"
Your role is to help the team connect to real data, move environments, and safely bring in external Agents—not to write SQL analysis or change the Harness source code.
What Lantide Data is (Platform Admin (Data Engineering) view)
In one sentence: a local desktop analysis IDE (Electron + local DuckDB) that can mount local files and external databases, where analysts do Plan → Execute → Report.
| It is | It is not |
|---|---|
| A local SQL query and Agent collaboration environment | A cloud ETL / scheduling platform |
| Able to ATTACH PostgreSQL / MySQL etc. for cross-source JOINs | A replacement for the company warehouse or BI |
Official .lantide bundles for project / workspace / settings migration |
"Seamless machine swap, import and go" (Rebind and reconnect required) |
What you should know: connections, workspace, data directory, .lantide export/import, plus an external Agent's scope, mode, and how to review it.
What you can skip: Series reading map §VI (engineering companions 3–4)—that Harness-contributor path, unless you're changing the product itself.
Typical data landscape (small and mid-size teams)
Data is often scattered across:
- Engineer-hosted PostgreSQL / MySQL
- Operations-owned CSV / Excel
Lantide uses DuckDB locally to chain both sides together in a single SQL statement, without having to ask someone to write ETL first. Your job is to point the connections and folders at the right place so analysts can get to work.
What you do in the flow
sequenceDiagram
participant Plat as Platform Admin (Data Engineering)
participant Analyst as Analyst
participant App as Lantide
Plat->>App: Configure connections and data directory
Plat->>App: Import a Workspace when needed
Plat->>App: Configure Agent Integration by trust level
Analyst->>App: Plan / Execute / Report
Plat->>App: Review Activity, Rotate or Revoke when needed
Plat->>App: Backup, machine-swap Rebind, reconnect DB
The analysis flow (Plan, Execute) is led by analysts and business / operations; you work in parallel on infrastructure—connections up, migration done, passwords and paths rebound.
Suggested reading path (onboarding → practice → advanced)
Onboarding: build a mental model (~12 minutes)
| # | Article | Overview | Status |
|---|---|---|---|
| 1 | Where Lantide sits in the data architecture | Local-first, not ETL | Available |
| 2 | Analyst vs Platform Admin (Data Engineering): who owns what | Connections / migration vs analysis delivery | Available |
Practice: day-to-day operations (~35 minutes + hands-on)
| # | Article | Overview | Status |
|---|---|---|---|
| 3 | External database connections | Test / Connect, Trial limits | Available |
| 4 | Workspace and data directory | Create, switch, isolate, change load dir | Available |
| 5 | .lantide export, import, and Rebind | Four-layer boundaries, machine-swap wizard | Available |
| — | Hands-on | One test connection + export and re-import a test Workspace | Available |
Advanced: optional and troubleshooting (~30 minutes)
| # | Article | Overview | Status |
|---|---|---|---|
| 6 | Safely connecting an external AI Agent to Lantide | Agent Integration's scope, mode, Activity, and revocation | Available |
| 7 | MCP Sources basics | Status and reconnect when Lantide connects to external data sources | Available |
| 8 | Backup, security boundaries, and common issues | Passwords not in the bundle, post-import checklist, connection troubleshooting | Available |
Design depth (optional)
| Resource | When to read |
|---|---|
| USER_GUIDE §3–4, §11.1 | Step-by-step operational cross-check |
| data-portability PRD | Bundle format and denylist detail |
| physical data PRD | Optional packing of local CSV/Parquet design |
Trial checklist
- Finish this article + the 2 onboarding articles
- For a test workspace, create a connection that completes Test Connection → Connect successfully
- Export a Workspace
.lantide, re-import it as a new workspace, and complete Rebind - (Optional) Drill Encrypt bundle or Include physical data files on a small bundle
- Confirm that after import you follow the prompts to Reconnect databases and, when needed, Refresh schema
- In the test workspace, drill an external Agent connection once with Observe, confirming the built-in Agent still works and Activity can be reviewed
- Share the Analyst introduction link with your analyst colleagues