Coming from…

Coming from Dataiku

If you build flows in Dataiku, Daedalus will feel close to home. You still draw a flow of steps from data to results. The difference is where it runs: on your own computer, with no server to set up, and your data stays there.

What feels familiar

The Flow is the canvas. You connect steps left to right, and you can see the whole workflow at a glance.
A recipe is a node. Group, Join, Stack, Pivot, Sort and Split each have a node of their own.
A dataset is a node's output. Select any node and its rows appear under the canvas, in the Data tab. Column types are in the Metadata tab.
Code recipes still exist. Write SQL in a DuckDB Query node. For Python, describe the step in words to a Code Gen node, or open any node's code with the pencil and edit it by hand.
Scenarios become scheduled runs. Deploy & schedule to cloud runs the workflow on a timetable and keeps a history of each run.
Dashboards become reports. The Reporting panel on the right holds your report layout, and a Write Report node turns it into a PDF, an HTML page or a PowerPoint deck.

What's different

A Prepare recipe is a chain of nodes. Each processor is its own node, so you can look at the data after every step.
No "build" step. Press Run and Daedalus recomputes the nodes that are out of date. Nodes you haven't changed keep their last result.
No machine-learning lab. There is Linear Regression and a Forecast node for time series, but no model training, scoring or evaluation screens.
A different set of connectors. Files, PostgreSQL, MySQL, SQLite, S3, Snowflake, Salesforce and web APIs are built in. Some databases Dataiku connects to are not, yet. See the list below.

Dataiku names in Daedalus

You don't need this table open while you work. Type a Dataiku name into the add-node menu (right-click the canvas) or the command palette (⌘K, Ctrl+K on Windows). The matching Daedalus node shows up with a small chip, such as Dataiku: Group, so you learn our name as you go.

Flow and running

Dataiku name	Daedalus
Build	The Run button (`⌘↵`), or Run pipeline in `⌘K`
Abort	Stop run in the toolbar or `⌘K`
Explore	The Data tab under the canvas
Schema	The Metadata tab under the canvas
Statistics	Describe, or the Metadata tab under the canvas
Charts	Chart, or the Chart tab under the canvas
Dashboards	Report Assembly or Write Report, or the Reporting panel on the right
Scenario	Deploy & schedule to cloud (Cloud runs)
Time-based trigger	Deploy & schedule to cloud (Cloud runs)
Version control	Version history
Workspaces	Share: a read-only link to the workflow
Generate Recipe	Extend with AI, which opens the Assistant pane
Prepare	Each Prepare step is its own node here; search the step's name

Connecting to data

Dataiku name	Daedalus
Upload your files	CSV Reader or Excel Reader
CSV	CSV Reader or Write CSV
Excel	Excel Reader or Write Excel
Parquet	Parquet Reader or Write Parquet
JSON	JSON Reader or Write JSON
Amazon S3	S3 Reader or S3 Sink
PostgreSQL	DB Reader or Write to DB
MySQL	DB Reader or Write to DB
API Connect	REST Fetch or HTTP Request
Editable dataset	Constant Table
ESRI Shapefiles	Geo Reader (drop the .shp file on the canvas) or Write Geo File

Visual recipes

Dataiku name	Daedalus
Sync	Write Parquet or Write to DB
Group	Aggregate or Group Aggregate
Join	Join or Lookup
Fuzzy join	Fuzzy Match
Geo join	Spatial Join
Stack	Union
Split	Switch or Random Split
Sort	Sort
Distinct	Distinct
Top N	Top / Bottom N or Row Limit
Pivot	Pivot
Window	Running Aggregations or Rolling Window
Sample/Filter	Filter or Random Sample
Export	Write CSV or Write Excel

Prepare processors

Dataiku name	Daedalus
Formula	Add Column or Multi-field Formula
Filter rows/cells on value	Filter
Remove rows where cell is empty	Filter or Fill Missing
Fill empty cells with fixed value	Fill Missing
Fill empty cells with previous/next value	Fill Missing
Rename columns	Rename
Delete/Keep columns by name	Select Columns
Split column	Split to Columns
Split and fold	Split to Rows
Fold multiple columns	Unpivot
Unfold	Pivot
Find and replace	Replace Values or Text Operations
Extract with regular expression	Regex Extract
Parse to standard date format	Parse or Cast Types
Extract date elements	Extract Date/Time Parts
Compute difference between dates	Date/Time Arithmetic
Format date with custom format	Convert Date/Time to Text
Discretize (bin) numerical values	Bin
Round numbers	Add Column or Multi-field Formula
Transform string	Text Operations
Split URL (into protocol, host, port, …)	URL Parse
Column Pseudonymization	Hash
Transpose rows to columns	Transpose

Code recipes

Dataiku name	Daedalus
Python	Code Gen or DuckDB Query
SQL query	DuckDB Query
SQL script	DuckDB Query
PySpark	Code Gen
SparkSQL	DuckDB Query
Prompt Recipe	No per-row LLM calls; Generate writes text per row from templates and rules

Data quality and statistics

Dataiku name	Daedalus
Checks	Assert or Expectations
Data quality rules	Expectations or Assert
Metrics	Describe or Value Counts
Generate statistics	Describe or Correlation Matrix
Correlation matrix	Correlation Matrix
Linear regression	Linear Regression

Geographic and text

Dataiku name	Daedalus
Change coordinates system	Reproject
Create GeoPoint from lat/lon	Make Point
Geocoder	Geocode
Reverse geocoding	Reverse Geocode
Create area around a geopoint	Buffer or Trade Areas
Isochrones	Isochrone
Geo Router	Route or Isochrone
OSM Enrichment	OSM Overpass or Overture Maps
Classify text	Classify
Sentiment Analysis	Classify
Named Entities Extraction	Extract
Summarize text	No LLM summaries; Generate builds text per row from templates and rules

Sending results

Dataiku name	Daedalus
Mail reporter	Email
Send message	Email or Slack
Webhook	HTTP Request

No equivalent yet

Some Dataiku tools have no Daedalus node yet. Searching for them shows a hint instead of an empty list. These are the ones you are most likely to miss:

Dataiku name	What to do instead
Train	No general model training yet; Linear Regression fits OLS and Forecast fits time series
Score	No model scoring; apply coefficients with Add Column
Evaluate	No model evaluation; compare with Add Column metrics
Cluster	No clustering yet; describe it to a Code Gen node
Principal Component Analysis	No PCA node yet; describe it to a Code Gen node
Text Embedding	No embeddings yet; describe it to a Code Gen node
Statistical tests	No test nodes yet; describe the test to a Code Gen node
Upsert	No upsert/merge write; Union + Distinct (keep last), then write
Excel Templater	No template filling; Write Excel writes plain named sheets
Shell	No shell steps; do the work in a Code Gen (Python) node
SparkR	No R support; rewrite the logic in a Code Gen (Python) node
Managed folders	Point readers at a project folder path instead
Google BigQuery	Not supported yet; export it to Parquet/CSV and read that file
Microsoft SQL Server	Not supported yet; export it to Parquet/CSV and read that file
Oracle	Not supported yet; export it to Parquet/CSV and read that file
Databricks	Not supported yet; export it to Parquet/CSV and read that file
Delta Lake	Not supported yet; export it to Parquet/CSV and read that file
SCP / SFTP	Not supported yet; download the file first, then read it
Flag holidays	No holiday calendar; Lookup a holiday table on the date

Tips

Search with the names you know. Typing Group, Stack or Fold multiple columns finds Aggregate, Union and Unpivot.
Rebuild a Prepare recipe one step at a time. Add a node per processor and check the Data tab after each one. A wrong step is easy to spot because every step has its own output.
Turn checks into a gate. Assert stops the run when the data breaks a rule. Expectations scores the data against the same kind of rules but never stops the run.
Before you schedule, check your sources. A cloud run has to reach every source. A database on your own network needs its drone made available to cloud runs first; the deploy dialog tells you which sources are ready.
Ask the Assistant. Describe the recipe in words, for example "group sales by region and keep the top 5", and it can build the nodes for you. You see each change it makes and can undo it.