Question

A legal-tech startup wants to build a solution that summarizes lengthy commercial contracts and extracts key clauses. The team has no ML engineers, wants to move quickly with minimal custom training, and needs to experiment with several foundation models before committing. They also want to deploy the chosen model to a managed endpoint on Google Cloud. Which approach best fits these requirements?

Accepted Answer

Use Vertex AI Model Garden to discover, test, and tune foundation models like Gemini, then deploy the selected model to a Vertex AI managed endpoint.

Answer

Train a custom transformer model from scratch in a Vertex AI custom training job using the startup's contract corpus.

Answer

Load all contracts into BigQuery and use BigQuery ML's linear regression to score clause importance.

Answer

Use Document AI's Invoice Parser to extract fields, then concatenate the fields to form a summary.