Did you know CSIRO’s Data61 open-source data catalog system, Magda https://magda.io/, powers data.gov.au and many other key infrastructure systems like national maps and digital twins across Australia — and we recently launched a new version of it with in-browser LLM integration? But our research and engineering team (Jacky Jiang Chen Wang et al.) has also been busy working on something much bigger: Magda++ — a powerful extension that turns your data catalog into an active discovery engine — much more than your RAG, simple question-answering about your data, and natural language to SQL/data analytics.
Magda++ is designed to handle exactly these challenges:
✅ Agentic Data Retrieval – Magda++ doesn’t just search your data — it understands it. It can intelligently query complex datasets, recognising relationships between variables and retrieving precise data points based on context, not just keywords.
✅ Dynamic Feature Engineering – Magda++ can shape your raw data into meaningful model features. For example, instead of just returning temperature data, it can calculate temperature anomalies, seasonal trends, and other derived features needed for predictive modelling.
✅ Counterfactual and “What-If” Analysis – What would happen if La Niña persists next season? Magda++ can simulate that scenario by adjusting key environmental variables (like rainfall and solar exposure) and predicting how chickpea protein yields might change — all while showing you which factors are driving the result.
✅ Explainable AI and Transparent Reasoning – Magda++ doesn’t just give you an answer — it shows you why it gave that answer. It traces how each variable influences the outcome, linking model behaviour back to real-world factors.

Magda++ is currently in testing mode with our partners and can integrate with other data catalogue systems. If you have a data catalogue system, we would be happy to discuss real use cases and gather your feedback.