# OuterProduct

## Docs

- [Aws Setup](https://docs.outerproduct.com/api-reference/account/aws-setup.md): Per-org IAM trust info the console renders for the connector setup snippet: OuterProduct's shared principal ARN + the org's stable ``external_id``.
- [Delete Account](https://docs.outerproduct.com/api-reference/account/delete-account.md): Delete the caller's auth user, cascade-clean org + keys.
- [Get User Session Api Key](https://docs.outerproduct.com/api-reference/account/get-user-session-api-key.md)
- [List User Session Api Keys](https://docs.outerproduct.com/api-reference/account/list-user-session-api-keys.md)
- [Me](https://docs.outerproduct.com/api-reference/account/me.md): Full profile for the console UI: identity + provider + the user's org row, hydrated from the small auth.users subset this API uses.
- [Mint User Session Api Key](https://docs.outerproduct.com/api-reference/account/mint-user-session-api-key.md): Mint a fresh API key for a verified user session.
- [Patch Org](https://docs.outerproduct.com/api-reference/account/patch-org.md)
- [Revoke Api Key](https://docs.outerproduct.com/api-reference/account/revoke-api-key.md): Soft-revoke (sets revoked_at). 404 if not found, already revoked, or owned by a different user.
- [Get Schema](https://docs.outerproduct.com/api-reference/agentic-documents/get-schema.md): Return the schema produced by a completed induce_schema job.
- [Get Table](https://docs.outerproduct.com/api-reference/agentic-documents/get-table.md): Return tabularize metadata for a completed job.
- [Induce Schema](https://docs.outerproduct.com/api-reference/agentic-documents/induce-schema.md): Submit an ``induce_schema`` job.
- [Tabularize](https://docs.outerproduct.com/api-reference/agentic-documents/tabularize.md): Submit a tabularize job.
- [Create Databricks Connector](https://docs.outerproduct.com/api-reference/connector/create-databricks-connector.md)
- [Create S3 Connector](https://docs.outerproduct.com/api-reference/connector/create-s3-connector.md)
- [Create Snowflake Connector](https://docs.outerproduct.com/api-reference/connector/create-snowflake-connector.md)
- [Delete Databricks Connector](https://docs.outerproduct.com/api-reference/connector/delete-databricks-connector.md)
- [Delete S3 Connector](https://docs.outerproduct.com/api-reference/connector/delete-s3-connector.md)
- [Delete Snowflake Connector](https://docs.outerproduct.com/api-reference/connector/delete-snowflake-connector.md)
- [List Connectors](https://docs.outerproduct.com/api-reference/connector/list-connectors.md)
- [Create Dataset](https://docs.outerproduct.com/api-reference/datasets/create-dataset.md): Persist a new dataset: the per-type connection_config row, then the ``datasets`` row referencing it.
- [Get Dataset](https://docs.outerproduct.com/api-reference/datasets/get-dataset.md)
- [List Datasets](https://docs.outerproduct.com/api-reference/datasets/list-datasets.md)
- [Explain](https://docs.outerproduct.com/api-reference/inference/explain.md): Per-sample local explanations.
- [Get Global Drivers](https://docs.outerproduct.com/api-reference/inference/get-global-drivers.md): Global feature importance.
- [Get Schema](https://docs.outerproduct.com/api-reference/inference/get-schema.md): Return the persisted schema manifest for a model.
- [Predict](https://docs.outerproduct.com/api-reference/inference/predict.md): Batch predictions from a trained model.
- [Predict And Explain](https://docs.outerproduct.com/api-reference/inference/predict-and-explain.md): Predict and explain in one call.
- [Scenario](https://docs.outerproduct.com/api-reference/inference/scenario.md): Counterfactual search: find row modifications that reach a desired class.
- [Get Job Result](https://docs.outerproduct.com/api-reference/jobs/get-job-result.md)
- [Get Status](https://docs.outerproduct.com/api-reference/jobs/get-status.md)
- [Get Model](https://docs.outerproduct.com/api-reference/models/get-model.md)
- [List Models](https://docs.outerproduct.com/api-reference/models/list-models.md)
- [Distribution Patterns](https://docs.outerproduct.com/api-reference/patterns/distribution-patterns.md): Per-pattern match rate over the supplied samples.
- [Fit Patterns](https://docs.outerproduct.com/api-reference/patterns/fit-patterns.md): Submit a pattern-tracker fit through the scheduler as a 1-node graph.
- [Get Pattern Tracker](https://docs.outerproduct.com/api-reference/patterns/get-pattern-tracker.md): Retrieve a fitted pattern tracker (or its current job status).
- [Partition Patterns](https://docs.outerproduct.com/api-reference/patterns/partition-patterns.md): Matching row indices (positional, into the request's ``samples``) keyed by pattern label.
- [Transform Patterns](https://docs.outerproduct.com/api-reference/patterns/transform-patterns.md): Evaluate every pattern against the supplied samples; return a ``(n_samples, n_patterns)`` boolean matrix aligned to ``labels``.
- [Health Check](https://docs.outerproduct.com/api-reference/system/health-check.md)
- [Version Info](https://docs.outerproduct.com/api-reference/system/version-info.md)
- [Reasoning Fit](https://docs.outerproduct.com/api-reference/training/reasoning-fit.md): Submit a reasoning fit and return immediately.
- [Trainer Run](https://docs.outerproduct.com/api-reference/training/trainer-run.md): Compile the trainer graph and hand it to the scheduler.
- [Create Document Upload](https://docs.outerproduct.com/api-reference/uploads/create-document-upload.md): Return a presigned URL for a single document.
- [Create Upload](https://docs.outerproduct.com/api-reference/uploads/create-upload.md): Mint a file_upload_connection_config row and return it with a presigned PUT.
- [Authenticate with OuterProduct: API Keys and op.init()](https://docs.outerproduct.com/authentication.md): Generate an OuterProduct API key and configure the Python SDK using an environment variable or by passing the key directly to op.init().
- [Connecting to Databricks](https://docs.outerproduct.com/guides/connectors/databricks.md): Reference Databricks SQL warehouse or cluster tables from OuterProduct using a stored access token. No Databricks credentials pass through the SDK.
- [Upload Local Files with LocalDataset](https://docs.outerproduct.com/guides/connectors/file-upload.md): Upload a local CSV or Parquet file to OuterProduct via a presigned URL with no console setup. Get a Dataset back and pass it straight to training.
- [Connect Your Data Sources](https://docs.outerproduct.com/guides/connectors/overview.md)
- [Amazon S3 with IAM Role Auth](https://docs.outerproduct.com/guides/connectors/s3.md): Reference files in your Amazon S3 bucket from OuterProduct using a stored IAM credential.
- [Snowflake with a PAT Credential](https://docs.outerproduct.com/guides/connectors/snowflake.md): Query Snowflake tables from OuterProduct using a stored Programmatic Access Token.
- [Counterfactual Scenarios: What Would Flip a Prediction?](https://docs.outerproduct.com/guides/counterfactuals.md): Use model.scenario() on a ReasoningModel to find minimal changes that flip a prediction to a target class, returned as structured ScenarioResult objects.
- [Feature Explanations](https://docs.outerproduct.com/guides/explanations.md): Use ReasoningModel to get per-sample feature attributions, human-readable decision rules, and global feature importance across your entire trained model.
- [Aggregate Model Prediction Patterns with PatternTracker](https://docs.outerproduct.com/guides/pattern-tracker.md): Fit a PatternTracker to distill a ReasoningModel's behavior on a prediction band into named, executable filter patterns you can apply to any new dataset.
- [Inference methods](https://docs.outerproduct.com/guides/predictions.md): Call model.predict() or model.predict_and_explain() on a trained Model or ReasoningModel to get a numpy.ndarray of predictions with schema validation.
- [Training models: HPO, Metrics, and Distillation](https://docs.outerproduct.com/guides/training.md): Train models on structured data using op.reasoning.fit() for a ReasoningModel or Trainer.configure().run() for our pure model training, with HPO and multi-metric optimization.
- [Model Lifecycle: A Complete Walkthrough](https://docs.outerproduct.com/guides/workflow.md): A complete walkthrough of the OuterProduct lifecycle: build a dataset, train a model, generate predictions, explanations, counterfactuals, and trackable patterns.
- [AI Reasoning for Structured Data](https://docs.outerproduct.com/introduction.md): OuterProduct offers a Python SDK and hosted API for AI reasoning on structured data.
- [Use OuterProduct from Claude, Cursor, and MCP Clients](https://docs.outerproduct.com/mcp/overview.md): Connect the OuterProduct MCP server to Claude, Cursor, or any MCP client and train models, surface reasoning, and run counterfactuals in plain English.
- [MCP Sample Prompts and Workflows](https://docs.outerproduct.com/mcp/prompts.md): Copy-paste prompts for Claude, Cursor, and other MCP clients that put OuterProduct's training, reasoning, and counterfactual tools to work immediately.
- [Quickstart: Train Reasoning Models](https://docs.outerproduct.com/quickstart.md): Install the OuterProduct SDK, connect your data, train a ReasoningModel, and generate predictions with feature-level reasoning in minutes.
- [outerproduct.agentic](https://docs.outerproduct.com/sdk/agentic.md): API reference for the agentic documents module — convert unstructured documents into typed structured features ready for ReasoningModel training.
- [outerproduct.connector](https://docs.outerproduct.com/sdk/connectors.md): API reference for S3Connector, SnowflakeConnector, and DatabricksConnector. For local files and in-memory frames, use LocalDataset.upload().
- [outerproduct.Dataset](https://docs.outerproduct.com/sdk/dataset.md): Stage local structured data with outerproduct.LocalDataset (CSV, Parquet, pandas, polars, NumPy) and upload it to a server-backed Dataset for training and inference.
- [outerproduct.init: Initialize the OuterProduct SDK](https://docs.outerproduct.com/sdk/init.md): Initialize the OuterProduct SDK with your API key. Must be called before any other SDK operation. Reads from environment or accepts a key directly.
- [outerproduct.model](https://docs.outerproduct.com/sdk/model.md): API reference for Model, ReasoningModel, ScenarioResult, and Predictor: the core objects returned by OuterProduct training and distillation flows.
- [outerproduct.reasoning](https://docs.outerproduct.com/sdk/reasoning.md): API reference for op.reasoning.fit() and PatternTracker: train reasoning models and distil population-level patterns from their predictions.
- [outerproduct.Trainer](https://docs.outerproduct.com/sdk/trainer.md): Train models on structured data with outerproduct.Trainer. Configure model families, optimization metrics, search strategy, and knowledge distillation.
- [Add Reasoning to a Recommender System](https://docs.outerproduct.com/use-cases/recommendations.md): Add feature-level reasoning to your approval-odds model to drive sign-up uplift, surface user cohort patterns, and deliver per-user counterfactual advice.
- [Build an Underwriting Reasoning App](https://docs.outerproduct.com/use-cases/underwriting.md): Bolt traceable reasoning onto your risk-scoring model for audit trails, policy optimization, and per-applicant counterfactuals.

## OpenAPI Specs

- [openapi](https://api.outerproduct.com/openapi.json)