Salesforce Docs Portal

Intelligent Context in Data 360

Process unstructured or structured data and create search index configurations tailored to specific business contexts by interacting with prompts in the Intelligent Context workspace.

Before you create a full search index, test your configuration on a small set of unstructured data. To start, add a few supported files or a data model object (DMO) in the Intelligent Context workspace. Respond to prompts to evaluate how the system processes your unstructured data and verify that you receive your expected results. Intelligent Context automatically saves your configuration, which then appears in the list.

After testing your configuration, apply it to an unstructured data model object (UDMO) or a DMO to create a precise, relevant search index.

Key benefits include:

  • Improved search accuracy: Give your teams and users quick access to accurate information by using LLM-powered content understanding, domain-specific intelligence, and user feedback loops.
  • Support for multiple file types: Upload a wide variety of file types, including DOCX, HTML, JPEG, JPG, PDF, PNG, PPTX, and TXT.
  • Structured data support: Select a data model object (DMO) as a source to build a precise search index directly from structured data, with no manual file exports required.
  • Customizable views: Gain multiple business perspectives from a single document. Design each lens to prioritize information critical to specific business functions so your teams can access relevant details without mining raw data.
  • Low-code setup: Use an intuitive interface to convert unstructured data into consumable search indexes without extensive technical expertise.
  • Efficient problem resolution: Diagnose and resolve common issues immediately, such as malformed tables in search results.
  • Harmonized citations: Harmonize your source content into a single, consistent schema to provide a uniform, renderable view across all source documents.

Intelligent Context Requirements

  • Enable Agentforce: Intelligent Context requires generative AI features available through Agentforce. If you haven't already, turn on Agentforce.
  • Turn On Einstein Generative AI: To use the generative AI features in Intelligent Context, you must turn on Einstein Generative AI in Data 360.

How to Use Intelligent Context

  • Validate content end-to-end: Before you use content in production systems, upload sample documents or select a DMO to verify that the search index locates critical information.
  • Customize prompts: Fix extraction issues, such as a malformed table, by adding simple prompts without rebuilding the entire search index. For example, "Extract the drivetrain table as markdown".
  • Metadata-driven refinement: Use SI Optimization to prepend metadata to generated chunks and apply pre-filter conditions, so searches return more precisely labeled, relevant results.
  • Deploy from test to production: Validate search quality on sample files, and then scale to thousands of documents by applying configurations to unstructured data model objects (UDMOs) or data model objects (DMOs).

Example: Adapting a Product Specification Document

Review how different teams can create distinct lenses to analyze a single product specification document.

  • Sales lens: Automatically extract information valuable to sales professionals, such as competitive advantages, pricing structures, and customer value propositions. This setup enables sales teams to identify and articulate product benefits quickly.
  • Service lens: Emphasize critical details for service and support teams, including support prerequisites, troubleshooting steps, and maintenance guidelines. This setup provides service teams with immediate access to information needed to resolve customer issues.