Salesforce hub — Data 360 families (7, 11, 12, 13) and the golden-record RAG pipeline for revenue (2026-10-23)

Re-maps the hub's families: Data 360 splits into foundation (7), unification and insights (11), retrieval and grounding (12) and automation and promotion (13), and revenue data and revenue agents (14) is added. Adds six Data 360 references, a TypeScript orchestrator that captures, applies, runs, waits for and verifies the pipeline, a guide to grounding a revenue agent, a linted example agent, and the capstone pattern from accounts, people and orders to an agent.

markdownreferencev1MWR2-4SQWarea: changelogchangelogrelease notesData 360 familiesgolden record RAG pipelinefamily map

Created 23 Oct 2026 · Modified 23 Oct 2026 · 1 version

Salesforce hub — Data 360 and the golden-record RAG pipeline (2026-10-23)

Family map

The old family 7 ("Data Cloud for grounding") was too broad, so it's split:

  • 7: Data 360 foundation;
  • 11: unification and insights;
  • 12: retrieval and grounding;
  • 13: automation and promotion;
  • 14 (new): revenue data and revenue agents, planned.

A second capstone, the golden-record RAG pipeline for revenue, is released. The agent-force demo rebuild is still planned. The map is reference/overview/families.md, and the charter has the same table.

Added

  • Family 7:
  • reference/data360/object-model-and-apis.md
  • reference/data360/streams-and-mapping.md
  • Family 11:
  • reference/data360/identity-resolution.md
  • reference/data360/insights-and-data-graphs.md
  • Family 12:
  • reference/data360/search-indexes-and-retrievers.md
  • guides/data360/ground-a-revenue-agent.md
  • examples/revenue-assistant-agent.md, which passes the Agent Script linter and compiler with no errors or warnings
  • Family 13:
  • reference/data360/automation-and-data-kits.md
  • guides/data360/automate-the-golden-record-pipeline.md, whose orchestrator compiles with tsc --strict
  • Capstone: patterns/golden-record-rag-pipeline-for-revenue.md
  • Tools: reference/overview/tools-ai-and-data360.md (tools split in two; Data 360 rows added)

Found while building

  • Salesforce publishes no CRM mapping for Order or OrderItem. Orders must be mapped to Sales Order and Sales Order Product by hand.
  • The Data 360 Connect API reference renders only in a browser, and older Connect REST pages render without request tables. The pipeline therefore captures definitions from a working org rather than guessing request bodies.
  • Salesforce's sf-skills contradicts itself on default_agent_user for employee agents that use the knowledge action.
  • There is no official CLI for Data 360 setup. Salesforce's own automation post uses Python, MuleSoft and Slack.

Trust

  • Official Salesforce sources (dated URLs and pinned commits):
  • object model and Sales Order DMO pages;
  • the data stream body (the February 2026 automation post);
  • the Query Connect API post;
  • calculated insight and identity resolution request fields;
  • data kit deploy payloads;
  • search index, retriever and Data Library steps;
  • sf-skills.
  • Labelled community in each document's provenance:
  • endpoint paths for runs, mappings, data graphs, search indexes and retrievers (the community sf-plugin-data360);
  • identity resolution limits and Account naming (a 2025 community guide).
  • Not verified in an org: request bodies beyond the documented fields, link-DMO column names, and the narrative-index approach. Each is marked "the hub's reading" or "capture".

Read this document as markdown — the same bytes an agent receives.