Studio hub
GAH2O.ai on the live graph. AutoML, Driverless AI, Document AI · models trained on graph entities, surfaced inline next to the things they describe.
What it is
Studio Models is the model-building workspace that lets you train, tune, interpret, score, deploy, and monitor machine learning models against interchangeable execution engines. It brings together automated training on tabular data, deep learning, document intelligence, retrieval-augmented chat, language model fine-tuning, and foundation-model predictions, all alongside guided setup wizards, an in-app AI copilot, real-time collaboration, and saved workspaces. Every model, deployment, prediction, evaluation, alert, and experiment is kept private to your workspace.
See it in the product
Top 4 capabilities
Each capability is the parent of dozens of typed features in the production taxonomy. Hover any feature in Studio to drill into the underlying nodes.
13 features
Tabular Model Training lets you train models on spreadsheet-style data using the full catalog of supervised and unsupervised algorithms, from linear models, boosted trees, and random forests to clustering, dimensionality reduction, anomaly detection, and ensembles. A fully automated mode does the work for you and returns a ranked leaderboard of the best-performing models.
11 features
Data Frames is where you prepare the tables that feed model training. Import data from files or databases, inspect its structure, run transformations, split it into training and validation sets, combine tables by row or column, derive new fields from text and dates, and export the results, all through validated, repeatable steps.
9 features
Document Intelligence pulls text and structure out of scanned documents, capturing text blocks, tables, key-value pairs, and signature regions for downstream training, annotation, or export. It handles multi-page files, preserves reading order, and scores its confidence on each region so low-confidence sections can be flagged for human review.
10 features
Data Model Reference is a built-in guide that maps each type of data in your workspace to how it is stored and related, so analysts and engineers can confirm where information lives and how it is shaped. It reads the live data model on demand, so the reference always reflects the current state.
More from the Studio hub
Composable dashboards built from graph queries. Refresh as the graph refreshes.
Whiteboard the workspace. Diagram strategy. Map the graph.
JupyterLab on the live graph. `pyramyd.query()` returns a DataFrame in one line · PYRAMYD widgets render inline · Y.Doc multiplayer with the workspace.
Spreadsheet-native views on the live graph. Edits push back to nodes.
Long-form documents co-authored with APEX and grounded in the graph.
Slide decks that refresh from the graph at present time.
30-minute walkthrough. We'll filter the workspace to your category and walk through the top capabilities live.
