Autonomous ETL.
Deterministic Results.
Let agents build and run your pipelines over MCP or the CLI, as a few readable rules your team reviews. Schedules and data-change triggers re-run them on their own, and every output row traces back to its rule and source data. Same input, same output.
clean_parts✓depends_on✓at_risk✓Proven in production
Imperative Code.
Or Declared Intent.
Most ETL cost is not compute. It is the code: written once, copied everywhere, understood by fewer people every year.
Say What.
Not How.
In Prometheux, a pipeline is a set of rules: connect your inputs, declare what each output means, and say where it goes. The engine handles joins, ordering and parallel execution, and runs recursive queries faster than optimized PySpark.
Connect, Define, Write
A connection reads the source, rules define the transformation, and an output writes the result. That is the whole pipeline.
SQL Where You Want It
Drop a SQL body into a rule when it reads better. The engine runs it alongside the rest of the pipeline.
Read and Write Anywhere
Read CSV, Parquet, Iceberg, JSON, Excel, relational databases and warehouses, and write to the ones that support it, with append or overwrite.
Reusable Definitions
Business concepts become reusable rules, kept as plain text in your own version control, so the next pipeline starts from them.
Ship new pipelines in days instead of sprints. Cut the code your team has to maintain. Trace any number back to the rule that made it.
A Pipeline in
Four Moves
Connect Your Inputs
Point Prometheux at a file, table or query. Credentials can sit in an external properties file.
Declare the Logic
Write rules for the joins, filters and derivations you need, in Vadalog or SQL.
Connect the Output
Choose where results land and whether to append or overwrite.
Run and Reuse
Run on demand or on a schedule; reuse the same rules in the next pipeline.
Less Code,
Faster Delivery
Regulatory Flows
Banca Sella runs a regulatory credit-risk flow on Prometheux rules, cutting it from 5 hours to under a minute.
Read the case study →Pipeline Consolidation
Replace bespoke ETL code with shared rules, lowering maintenance and the risk of losing knowledge when teams change.
Consistent Metrics
Define a KPI once as a rule and feed every report and dashboard from the same definition.
Training-Ready Data
Use the same declarative rules to prepare clean, consistent datasets for models and agents.
“Prometheux enabled us to build a system that can access our databases, process all the client's historical variables, and provide an intelligent, understandable explanation autonomously.”
Massimiliano MalinHead of Credit Risk · Banca Sella
“Generative AI without ontologies is an engine without a road. Prometheux builds the road.”
Marco RomeiDeputy CIO · Banca Sella
Questions,
Answered
What does 'declarative ETL' mean here?
You state what each output is, as rules over your inputs. Prometheux works out how to join, filter and execute it, instead of you writing the steps by hand.
How is this different from SQL transformation tools?
Those bring version control and testing to SQL that runs inside one warehouse. Prometheux rules are also plain text you version, but one program reads from and writes to many sources, recurses through hierarchies to any depth, and links every result to the rule and source data that produced it.
Do I have to learn a new language?
No. Write rules in plain language and let the AI assistant draft them for you to review, or write them in SQL, Cypher or Python. Vadalog, a readable declarative logic language, is there when you want full control, and every option shares the same concepts and lineage.
Can business users check the logic?
Yes. A rule states what an output means in a line or two, and the assistant can translate any rule into plain language. Execution doesn't depend on a model: the same input always gives the same output.
Which sources and targets can a pipeline use?
Pipelines read files (CSV, Parquet, Iceberg, JSON, Excel), relational databases such as PostgreSQL, Oracle and SQL Server, and warehouses such as Snowflake, Databricks, BigQuery and Redshift. They write to CSV, Parquet, Iceberg, Excel, PostgreSQL, SQL Server, Snowflake, Databricks and more; a few sources, such as Oracle and BigQuery, are read-only.
Does it replace my orchestrator?
No. Prometheux defines and runs the transformation logic. Concepts can run on a schedule or when source data changes, write results to your targets, or be triggered from the orchestrator you already use through the REST API, Python SDK or CLI.
Related Capabilities
Shrink Your
Next Pipeline
Bring a PySpark or SQL job you'd rather not maintain. We'll rewrite it as rules with you.
