Skip to content
PlaidCloud

Start From an Existing Project

~15 minutes · copy or import a project, then make it yours

The fastest day-1 win in PlaidCloud is not building from nothing — it’s starting from something that already works. Pick a project that already imports data, transforms it, and produces a result, get your own copy of it, and change the parts that are yours.

There is no template gallery to browse. Instead you have two grounded ways to begin from a working project:

  • Copy a project that already exists in your workspace into a brand-new project of your own — Copying a Project.
  • Run a portable bundle — a project someone exported as a self-contained portable-format package, which runs on its own outside PlaidCloud.

Which one you want depends on where you’re headed.

a working projectPath A · Copy Projecta new project you own, in PlaidCloudPath B · Portable Bundlea zip that runs off-platformstays inside PlaidCloud · edit and run as normalPython + DuckDB · one-way snapshot
Copy to keep working in PlaidCloud; export a portable bundle to take a snapshot off-platform.
You want to… Use Result
Build on a project inside PlaidCloud and keep iterating Copy Project A new project you own, ready to edit and run
Run a project’s logic offline, or hand it to someone with no PlaidCloud login Export to Portable Format A standalone Python + DuckDB package

Prerequisites

Section titled “Prerequisites”
  • A PlaidCloud workspace (start a free trial if you don’t have one).
  • For Path A: an existing project you can see in the Projects tab — one of your own, a shared sample, or anything a colleague has given you access to.
  • For Path B: a portable-format .zip someone has exported, or a project of your own to export.

Not sure what a project is? A project is the unit of work in PlaidCloud — it owns its own data, workflows, dimensions, and history, and doesn’t share state with any other project. Read Projects first if this is new.

Path A: Copy an Existing Project

Section titled “Path A: Copy an Existing Project”

Copying makes a full, editable duplicate — all of it, or just the parts you pick — into a brand-new project that’s entirely yours.

  1. Open Analyze and select the Projects tab.
  2. Select the project you want to start from.
  3. Click the Actions button and choose Copy Project.
  4. Choose a new project as the target. Give it a name and an ID.
  5. Leave the selection tree fully ticked to bring the whole project across, then click Copy Project.

The ID is permanent. The ID you give the new project becomes its schema name and cannot be changed later. Pick something you’re happy to live with — the name is free to change any time, the ID is not.

The copy runs in the background; you’re notified when it finishes. See Copying a Project for the full dialog.

Bring Only the Parts You Want

Section titled “Bring Only the Parts You Want”

You don’t have to copy everything. The selection tree lists what the project holds by type — workflows, steps, tables, dimensions, data editors, user-defined functions — and you can tick only the parts you want. Anything you tick brings its dependencies with it automatically: the tables a step reads and writes, the dimensions it loads against, the functions it calls, and any sub-workflow it invokes. You never have to work out those dependencies yourself.

Keep references pointing at your copy. Leave Update References on (its default). It rewrites the copied steps so they point at the tables and objects in your new project rather than back at the original — almost always what you want. Turn it off only if you deliberately want the copy to keep reading the source project’s objects.

Path B: Start From a Portable Bundle

Section titled “Path B: Start From a Portable Bundle”

A portable bundle is a project packaged as a self-contained Python program that runs on DuckDB with no connection to PlaidCloud. It’s the way to start from a working project when you want it to run off-platform — on your laptop, in a CI job, or in the hands of someone who has no PlaidCloud login.

If a colleague has already given you a bundle, skip to Run It. To make one yourself:

  1. Open Analyze and select the Projects tab.
  2. Select a single project.
  3. Open the Actions menu and choose Export to Portable Format.
  4. Set the Document Account and Folder Path the package is written to, leave Include data checked so it runs without any external source, and click Export.

The result is a .zip containing one Python script per workflow, the Parquet data snapshots, a requirements.txt listing the Python packages the scripts need, and a README.md, COVERAGE.md, and RUN_ORDER.md. See Export to Portable Format for what each file is.

A portable bundle is a one-way snapshot, not a re-import. It runs outside PlaidCloud and does not sync back — changes you make to the package don’t return to the project, and re-exporting after edits produces a fresh package. To keep working inside PlaidCloud, use Path A instead. To keep iterating in PlaidCloud and hand out a runnable copy, do both.

Fidelity is best-effort. SQL transform steps are reproduced exactly. Proprietary and side-effecting steps — allocations, dimension loads, notifications, document and report steps, scripts — are exported as stubs that don’t re-run; their last materialized output is bundled as a snapshot instead. Each step’s fidelity is recorded in COVERAGE.md. For fully recomputed results, run the project in PlaidCloud.

Run It

Section titled “Run It”

Unzip the package, create a Python environment, and run each workflow script in the order RUN_ORDER.md lists:

Terminal window
python -m venv .venv && source .venv/bin/activate
pip install -r requirements.txt
python pipeline_<workflow>.py

Each script builds its tables in a local DuckDB database. SQL steps recompute live; stubbed steps leave their snapshotted output in place.

Make It Yours

Section titled “Make It Yours”

However you started, the copy is now yours to change — this is where the day-1 win pays off. A few first moves:

Do this Where
Point imports at your data instead of the sample’s The import steps in each workflow — see Load, Transform, and Publish Data
Rename the project or adjust who can open it Managing Projects → Access Control
Set project variables for values that recur across workflows Managing Project Variables
Confirm the whole thing still runs end to end Click Run on each workflow and watch the log

Because every change is tracked, you can experiment freely — automatic change tracking lets you recover any earlier point-in-time state if an edit goes wrong.

Where to Go Next

Section titled “Where to Go Next”