Salesforce & AI Glossary

Key terms for enterprise Salesforce architecture — from Agentforce and Data Cloud to technical debt and org health. Each definition written for architects and decision-makers, not marketing decks.

A

Agentforce

Salesforce's AI agent platform that enables autonomous digital workers to handle customer interactions, process requests, and execute business logic across Sales, Service, and Experience Cloud. Related service →

Atlas Reasoning Engine

The large language model reasoning layer powering Agentforce agents. It processes user inputs, evaluates topic models, selects appropriate actions, and generates contextual responses based on available data and guardrails. Related service →

C

Calculated Insights

Derived metrics in Data Cloud computed from unified profile data — such as lifetime value, engagement scores, or propensity models — that feed downstream segmentation and activation. Related service →

Center of Excellence (CoE)

A cross-functional governance body that sets standards, manages the backlog, and coordinates releases across Salesforce clouds. The difference between a CoE that works and one that stalls is decision rights clarity. Related service →

CRM Analytics

Salesforce's embedded analytics platform (formerly Tableau CRM / Einstein Analytics) for building dashboards, lenses, and predictive models directly on CRM data without ETL.

Customer 360

Salesforce's vision of a single, unified customer profile across all clouds and touchpoints. In practice, achieving it requires identity resolution, data harmonization, and cross-cloud integration — not just licensing. Related service →

D

Data Cloud

Salesforce's customer data platform that ingests, harmonizes, and activates data from any source. It unifies customer identity through probabilistic and deterministic matching, enabling real-time segmentation and activation. Related service →

Data Graphs

Structured, queryable representations of related Data Model Objects in Data Cloud. They define which objects and relationships are accessible to downstream consumers like Agentforce agents or Prompt Builder templates. Related service →

Data Model Objects (DMOs)

The canonical schema in Data Cloud that maps ingested data to a standardized model. DMOs bridge the gap between raw source data and the unified profile used for segmentation and activation. Related service →

Data Streams

Ingestion connectors in Data Cloud that pull data from external sources — Salesforce CRM, MuleSoft, S3, APIs, or file uploads — into the lakehouse for harmonization. Related service →

E

External Services

A Salesforce feature that registers external REST APIs as declarative actions, making them invocable from Flows, Agentforce agents, and Apex without custom integration code. Related service →

G

Governor Limits

Salesforce's hard runtime constraints on CPU time, SOQL queries, DML operations, and heap size per transaction. Hitting them in production usually signals architectural debt — not a coding mistake. Related service →

I

Identity Resolution

The Data Cloud process that matches and merges records from multiple sources into a single Unified Individual. Uses deterministic rules (exact email match) and probabilistic rules (fuzzy matching on name + address). Related service →

M

MuleSoft

Salesforce's API-led integration platform for connecting systems, transforming data, and orchestrating business processes. It moves data between systems — a different job than Data Cloud, which unifies identity inside Salesforce. Related service →

Multi-Cloud Architecture

A Salesforce deployment pattern spanning two or more clouds (Sales, Service, Marketing, Commerce, Experience) with shared data, unified identity, and cross-cloud automation. The complexity is in the seams, not the clouds. Related service →

O

Org Health Assessment

A systematic audit of a Salesforce org covering technical debt, automation sprawl, permission complexity, data model integrity, and deployment practices. The output is a prioritized remediation roadmap. Related service →

P

Platform Events

Salesforce's event-driven messaging system for publishing and subscribing to real-time events across clouds and external systems. Used for decoupling integrations and triggering asynchronous processes.

Prompt Builder

A Salesforce tool for creating reusable prompt templates that ground LLM responses in CRM data. Templates can merge fields from records, Data Cloud profiles, and Flow outputs into structured prompts. Related service →

S

Salesforce Flow

The declarative automation engine in Salesforce for building record-triggered, screen, scheduled, and platform event-triggered processes. When Flows proliferate without governance, they become the top source of org health debt. Related service →

Segments

Audience definitions in Data Cloud built from unified profile attributes, calculated insights, and behavioral data. Segments activate to downstream targets — Marketing Cloud, Ads, or external systems — for personalized engagement. Related service →

Solution Architecture

The discipline of designing end-to-end technical solutions across Salesforce clouds, integrations, and data layers. A solution architect owns the blueprint — not the backlog, not the sprint, not the code.

Staff Augmentation (Régie)

An engagement model where a freelance architect embeds within the client's team for a defined period. Common in French SI contracts as "régie" — the architect works on-site under the client's direction.

T

Technical Debt

The accumulated cost of architectural shortcuts, undocumented customizations, and deferred maintenance in a Salesforce org. It compounds: every new feature built on top of debt costs more than it should. Related service →

U

Unified Individual

The resolved, deduplicated customer profile in Data Cloud after identity resolution. It merges data from CRM records, web interactions, support tickets, and external sources into a single canonical identity. Related service →

Z

Zero-Copy Data Federation

A Data Cloud capability that queries external data warehouses (Snowflake, BigQuery, Databricks) in place — without copying or moving the data. Reduces data duplication and latency for large-scale analytics. Related service →