Use tools in Data 360 to process unstructured content and structured data for use in AI, automation, and analytics workflows.
- Enterprise Knowledge for Agentforce Service
Use Enterprise Knowledge to unify your siloed content. With Enterprise Knowledge you can ingest content into Data 360, harmonize it into a consistent harmonized data model object (HDMO), and gain valuable insights on your user activity. Once the content is in Enterprise Knowledge, use it to power your AI agent’s decisions and Service Agent case handling. - Process More File Types in Intelligent Context
Intelligent Context allows a variety of file types to create an accurate search index. Upload the following file types to any Intelligent Context workspace: .docx, .html, .jpeg, .pdf, .png, .pptx, .txt. Then, create and publish your search index for improved search accuracy. - Filter and Enhance Local File Uploads with Metadata
You can now add, map, and use metadata fields such as subject and file type when uploading local files. Use these fields to apply prepend values directly to the generated document chunks, or set pre-filter conditions to refine query. Gain finer control over how your unstructured data is chunked and searched. - Use Structured Data Model Objects as Sources
Select a structured data model object (DMO) as a source to build your Intelligent Context configurations. To save and create the index, select up to 30 records from the DMO at a time. - Process More Pages and Larger Files in Document AI
You can now process up to 100 pages (increased from 25) totaling 25 MB (increased from 10 MB) in batch processing mode. - Process Documents with Additional Large Language Models in Document AI
Get higher extraction accuracy and faster processing for complex documents with the latest large language models (LLMs). Choose from Gemini 3 Flash, GPT 5.4, or Claude Sonnet 4.6. - Refine a Document AI Configuration with Drafts (Generally Available)
Refine your Document AI extraction schema across multiple work sessions by saving each iteration as a draft. You can continue to edit prompts and test the draft before publishing to production. - Export and Import Document AI Schemas as JSON (Generally Available)
Export and reuse Document AI extraction schemas across projects and teams so you don't have to manually recreate a configuration. Export a JSON schema from a Document AI configuration, and import and edit the JSON schema file to create a new configuration. - Extract Data from Word, PowerPoint, and Excel Files in Document AI
Extract structured data from Microsoft Office files with the Connect API. Pass Word (.docx), PowerPoint (.pptx), and single-sheet Excel (.xlsx) files without requiring any file conversion or rigid cell-mapping. - Enrich Records from Documents with SmartFill
Populate record fields directly by uploading documents in various file formats. Upload a sample file, review and refine the field mappings, and add the Edit with SmartFill action to a record page. Your users upload files in SmartFill to fill forms instead of entering data manually. - Process Word and PowerPoint Files Natively in Document AI Batch Jobs
Use Document AI to process Microsoft Word (.docx) and PowerPoint (.pptx) files in batch jobs. Extract structured data from your high-volume documents, such as offer letters, pitch decks, and legal contracts. Process files up to 10 MB or 25 pages without converting them to PDF first. - Process Unstructured Data with Your Preferred LLM
Parse and preprocess unstructured data using any supported large language model (LLM). Previously, you could use only GPT-4o for LLM-based Parsing and LLM-based Visual preprocessing. Now, you can override the default GPT-4o selection to use the LLM that best fits your business needs and your content. - Process Complex Tables with Docling Parser and LLM-based Image Processing
Enhance retrieval augmented generation (RAG) responses with information extracted from complex table layouts, such as tables with merged cells, nested headers, or multi-page table splits. When you select the Docling parser with LLM-based image processing for documents with complex tables, Data 360 chunks the information optimally. This process preserves the context throughout and reduces hallucinations. - Use LLM-Based Parsing and Preprocessing on DMOs with File Attachments
Process Data 360 data model objects (DMOs) that include a file attachment by selecting LLM-based parsing and preprocessing options. You can also view the process history metrics for these search indexes. Previously, LLM-based parsing and preprocessing options and process history metrics weren’t available for search indexes built with file attachment unstructured data model objects (UDMOs). - Preserve Agent Memory Across Sessions Without Requiring Customers to Repeat Information
Provide personalized agent interactions where every conversation builds on what came before without requiring users to repeat themselves across sessions or agent handoffs. Register your agents in Data 360 to enable memory capabilities. Data 360 automatically extracts facts, preferences, and conversation summaries from interactions and enforces your organization's access controls when delivering context to your agents. - Register External Agents Without API Calls
Set up external agents for memory extraction and context without writing API code. Use a new form in Agent Config to register agents by entering basic details. Data 360 provisions the data graphs, DLOs, DMOs, connectors, and search indexes for your agents to access curated context. - Keep User Profiles Current Without Manual Updates
Equip your agents to automatically extract facts and preferences from conversations so they can personalize future interactions. Data 360 extracts this information in real time as it ingests conversations, with no configuration, rules, or prompts required. Data 360 updates facts automatically as users mention new details or change their preferences, and resolves conflicts in favor of the most recent statement. - Give Agents a Complete Picture of Every User Across Channels
Unify conversation history and memories for each user, known or anonymous, across channels and devices, so your agents always work from a complete picture of who they're talking to. Whether a customer reaches out by chat, phone, or another known or anonymous channel, Data 360 connects those interactions to the same person. Agents pick up with full context, no matter which channel or device the user comes from.