Process Content in Data Cloud

Use generative AI-based tools in Data Cloud to process unstructured data for use in AI, automation, and analytics workflows.

Processing capabilities available in Data Cloud include:

• Prompt-based customization: Use custom prompts or prompts powered by agents to chunk and further process your unstructured data for search index optimization. You can also use prompts to fix extraction issues, such as malformed tables, without rebuilding the entire search index.
• Indexed audio and video transcriptions: Simplify search and retrieval of audio from audio and video data by transcribing and indexing content automatically in Data Cloud. You can then query transcriptions to retrieve relevant information or similar items based on the text.
• Enriched indexes: Use AI to extract metadata from indexed content. Extracted metadata includes keywords, entities, topic overviews, questions answered by the content, and content summaries.
• Content parsing and preprocessing: Use prompts to identify document structure and convert text and metadata from unstructured documents into structured or semi-structured representations (parsing). Or ensure that data extracted from images and tables maintains the context and relationships from the source (preprocessing).
• Document schema extraction: Generate structured output from unstructured content. Define schemas to extract, classify, or generate fields and tables using prompts or programmatically with the Data Cloud Connect API.

Consider these key use cases for content processing in Data Cloud.

• For Retrieval Augmented Generation (RAG): By processing unstructured data and adding your content to a Data Cloud search index, you enable AI to retrieve precise context and ground its responses in your specific data for more accurate and relevant outputs. This is the foundation of RAG.
• For Service Cloud: Content processing extracts key information and topics from knowledge articles, emails, support tickets, and chat logs. This structured output allows AI agents to find precise answers, craft better responses, and proactively suggest solutions to customer issues.
• For Sales Cloud: Processing unstructured sales data, such as emails, call transcripts, and notes, extracts key details and sentiment. AI agents use this information to generate highly relevant meeting briefings and summaries, and to fuel personalized content recommendations that boost engagement.
• For automation flows: Content processing provides the semantic understanding needed to compare records, such as cases, based on their descriptions. This allows AI agents to automatically identify duplicates, accelerate resolution times by finding similar historical examples, and accurately classify incoming records.
• For analytics: Processing unstructured data prepares it for deep analysis in Tableau. By extracting themes and classifying topics within your content, you can reveal patterns and gain insights that are often missed by traditional analytics methods.

Process content in Data Cloud using these tools.

• Unstructured Data
  Use unstructured data in Data 360 to ground Agentforce agents, generative AI, analytics, and automation use cases with business-specific data that delivers deeper insights for your users and customers.
• Chunk Data
  To add your data to Data 360’s search index, you must first chunk it. Chunking break your data into meaningful chunks, and Data 360 turns those chunks into machine-readable vector embeddings.
• Intelligent Context
  Intelligent Context in Data 360 is a hands-on, AI-powered workspace where you can interact with prompts to process unstructured data to create, modify, test, and publish search index configurations tailored to specific business contexts.
• Document AI
  Use Document AI in Data 360 to extract structured data from unstructured documents like invoices, resumes, lab reports, and purchase orders.
• Parsing and Preprocessing Content
  Prepare unstructured data for chunking by using Data 360 content parsing and preprocessing methods. Parsing and preprocessing identify document structures, such as headings, paragraphs, or tables, remove irrelevant content, and prepare the text for chunking and indexing.
• Search Index
  Data Cloud uses a search index to manage structured and unstructured content in a search-optimized way. Create a search index configuration to chunk and vectorize yout content. Chunking breaks the text into smaller units, reflecting passages of the original content, such as sentences or paragraphs. Vectorization converts chunks into numeric representations of the text that capture semantic similarities.