AIki
The AI + Salesforce Glossary 98 terms and counting
A2A (Agent-to-Agent Protocol)
Google's open protocol that allows AI agents from different vendors to communicate and collaborate with each other.
A2UI (Agent-to-UI Protocol)
Google's open protocol that lets AI agents generate safe, interactive UI components — forms, cards, pickers — from a pre-approved catalog, instead of responding with text.
AG-UI (Agent-User Interaction Protocol)
An open protocol that streams an AI agent's actions, state, and output into the user's frontend in real time.
Agent Builder
Salesforce's low-code tool for creating and configuring Agentforce agents — define topics, actions, and guardrails without writing code.
Agent Commerce
The emerging market category where AI agents transact directly — paying for APIs, buying premium content, hiring other agents, and settling between machines without human approval per transaction. The economic activity layer above the agent stack.
Agent Control Plane
The centralized layer that governs running agents — defining their identity, scopes, tool access, policies, and lifecycle — separate from the data plane where the agents actually execute.
Agent Development Life Cycle (ADLC)
A structured methodology for planning, building, testing, deploying, and maintaining AI agents in enterprise environments, adapting SDLC practices for agent-specific concerns.
Agent Governance
The policies, controls, and monitoring systems that ensure AI agents operate safely, compliantly, and within business-approved boundaries.
Agent Handoff Protocol
The structured rules governing how one AI agent transfers context and control to another agent or human.
Agent Infrastructure
The runtime, network, and tooling substrate that AI agents need to execute reliably — sandboxed compute, tool access, memory, gateways to LLM providers, and the orchestration plumbing that connects them. Closer to the metal than agent operations.
Agent Memory
The systems that let an AI agent retain context across calls, sessions, users, or teams — turning a stateless model into something with continuity. Encompasses short-term working memory, long-term episodic memory, and shared organizational memory.
Agent Observability
The practice of inspecting, debugging, and understanding AI agent behavior at runtime by consuming agent telemetry — traces, metrics, logs, and events — through dashboards, alerts, and evaluation tools.
Agent Operations
The discipline of running AI agents in production — capturing what they do, attributing what it costs, evaluating what they produce, and intervening when something goes wrong. The operational layer above agent observability and orchestration.
Agent Orchestration
The coordination and management of multiple AI agents working together to accomplish complex workflows that no single agent could handle alone.
Agent Telemetry
The runtime data emitted by an AI agent — every decision, tool call, input, output, latency, and cost — used to monitor reliability, quality, and spend in production.
Agentforce
Salesforce's AI agent platform that enables businesses to build, customize, and deploy autonomous AI agents across sales, service, marketing, and commerce.
Agentforce 360
Salesforce's unified AI platform announced at Dreamforce 2025, bringing together Agentforce agents, Data 360, Agentforce Voice, and cloud-specific AI capabilities under one umbrella.
Agentic AI
AI systems designed to take autonomous action, not just generate content or answer questions. The shift from "AI that talks" to "AI that does."
Agentic Architecture
The design pattern for building AI systems that plan, act through tools, remember, and stay under guardrails and human approval gates.
Agentic Enterprise
An organization that has shifted its core business processes from manual workflows and traditional software to autonomous AI agents as the primary operating system.
AgentOps
Shorthand for the practice of running AI agents in production — borrowed from "DevOps" and "MLOps" — encompassing observability, cost attribution, evaluation, and the operational discipline of managing agents at scale. Often used interchangeably with "agent operations."
AI
AI, or artificial intelligence, refers to the simulation of human intelligence processes by machines, particularly computer systems.
AI Agent
An autonomous AI system that can perceive its environment, make decisions, and take actions to achieve specific goals - without constant human direction.
AI OS
A loosely defined category for systems that position AI — typically an LLM-driven assistant or agent layer — as the primary interface between the user and computing, the way Windows or iOS sit between the user and the underlying machine. Used in three distinct ways depending on the speaker.
AI Readiness Assessment
A structured evaluation of whether your Salesforce org, data, processes, and team are prepared to deploy AI agents successfully.
AI-Native Application
Software architected around an AI model and agents from the first line of code, rather than a traditional app with an AI feature added later.
Apex
Salesforce's proprietary programming language for custom business logic — the backend code behind your automations, integrations, and agent actions.
Atlas Reasoning Engine
The AI brain inside Agentforce that plans multi-step actions, evaluates options, and decides what to do next — all grounded in your Salesforce data.
Autonomous Agent
An AI agent that operates independently, making decisions and taking actions with minimal or no human oversight, within predefined boundaries.
Capability Registry
A structured catalog that maps AI capabilities (reasoning, structured output, tool use, vision, long context) to the models that can serve them — the substrate that makes skills portable across LLM vendors.
Capability-Based Authorization
Authorizing an AI agent by granting narrow, revocable permissions for specific actions, rather than handing it a broad API key or a human's full access.
Cascading Hallucinations
When one agent's hallucinated output is consumed as truth by the next agent in a workflow — quietly spreading bad data across systems and decisions before any human sees it.
Claudeforce
The Salesforce–Anthropic partnership announced 26 August 2026: Claude inside Agentforce and Slack, and Salesforce inside Claude.
Company Brain
A company-wide AI knowledge system that indexes every internal artifact — docs, email, Slack, meetings, CRM, tickets — and exposes it as a queryable, agent-accessible memory. The organizational equivalent of a "second brain," built on RAG plus connectors plus agents.
Copilot Actions
Pre-built AI capabilities in Salesforce that handle common tasks like summarizing records, drafting emails, and generating reports — no configuration required.
Customer 360
Salesforce's vision (and product suite) for having a complete, unified view of every customer across all touchpoints - sales, service, marketing, commerce.
Customer Relationship Management
The technology and discipline for managing every customer interaction across sales, service, and marketing — keeping a single source of truth for every account, contact, and conversation.
Data 360 (formerly Data Cloud)
Salesforce's real-time data platform that unifies customer data from any source into a single customer profile. Rebranded from Data Cloud at Dreamforce 2025.
Data Cloud
Salesforce's unified data platform that centralizes customer data from all sources, making it available for agents, analytics, and insights in real time.
Digital Labor
AI agents and automation performing work traditionally done by human employees, measurable in tasks and hours saved.
Einstein
Salesforce's AI brand, originally launched in 2016, covering predictive analytics, recommendations, and now generative AI features.
Einstein Copilot
Salesforce's AI assistant built into the Salesforce platform, providing real-time guidance and automation across CRM workflows.
Embedding
A numerical representation of text that captures its meaning — the technology that powers semantic search, RAG, and AI-driven recommendations.
Enterprise AI
Enterprise AI refers to the integration of artificial intelligence technologies into larger business operations and processes.
Experience Cloud
Salesforce's platform for building customer portals, partner communities, and self-service sites — the digital front door for external users.
Field-Level Security (FLS)
Salesforce's per-field read and edit permission control — the mechanism that keeps an Agentforce agent from seeing or changing fields it shouldn't.
Flex Credits
Salesforce's consumption-based pricing unit for Agentforce — you pay per conversation, not per user license.
Foundation Model
A large AI model trained on broad data at massive scale that serves as a general-purpose substrate other applications fine-tune or prompt against. The category that includes GPT, Claude, Gemini, Grok, Llama, and the rest of the frontier model lineup.
GraphQL
A query language for APIs that lets a client ask for exactly the fields it needs in one request — the data access layer for Salesforce Multi-Framework apps.
Grounding
Connecting AI responses to your actual business data so agents answer with facts, not hallucinations.
Guardrails
The boundaries you set on AI agents to control what they can discuss, what actions they can take, and when they must escalate to a human.
Headless 360
Salesforce’s architectural shift exposing every platform capability as an API, MCP tool or CLI command — making the browser interface optional so agents can plug in natively.
Headless Architecture
A software design pattern that separates backend logic and data from the presentation layer, letting any frontend — UI, API, agent, or CLI — consume the same business capabilities.
Human-in-the-Loop (HITL)
A design pattern where AI systems require human approval or intervention at critical decision points before taking action.
Human-in-the-Loop Overwhelm
A social-engineering attack on the humans approving agent actions — flooding the review queue until reviewers rubber-stamp risky requests out of fatigue.
Hyperforce
Salesforce's cloud infrastructure that replaced its own data centres, and the org requirement you must check before scoping any Multi-Framework React build.
Idempotency
A property ensuring an action produces the same result no matter how many times it's performed.
Inference Chips
Specialized silicon designed to run trained AI models (the "inference" step) rather than train them — typically optimized for low latency, high tokens-per-second, and lower cost-per-token than the general-purpose GPUs used for training.
L402 Protocol
A machine-to-machine payment protocol that combines HTTP status code 402 ("Payment Required") with the Bitcoin Lightning Network, designed for AI agents and APIs to transact in fractions-of-a-cent without involving credit cards or human approval. Originally called LSAT by Lightning Labs.
LLM (Large Language Model)
The AI technology behind ChatGPT, Claude, and the intelligence in Agentforce. Trained on massive amounts of text to understand and generate human language.
LLM Cost Attribution
The practice of tying every LLM call back to the task, agent, process, or skill that triggered it — across every vendor — so AI spend can be measured against outcomes, not just tokens.
LLM Gateway
A unified proxy in front of multiple LLM providers that captures every call, enforces policy, and lets a single application talk to Anthropic, OpenAI, xAI, Gemini, and local models through one interface.
Machine-Payable APIs
APIs that expose their price natively in their HTTP response and accept payment from a calling client without human involvement — the substrate that makes agent commerce possible. Typically implemented over HTTP 402 plus a payment protocol like L402 or x402.
MCP (Model Context Protocol)
Anthropic's open standard for connecting AI models to external data sources and tools. Think of it as a universal adapter for AI.
MCP-UI (MCP Apps Extension)
An official extension to the Model Context Protocol that lets MCP tools return interactive UI components, not just text or data.
MELT (Metrics, Events, Logs, Traces)
The four-pillar framework for organizing observability data: Metrics (numeric signals), Events (state transitions), Logs (per-step context), and Traces (end-to-end request chains).
Micro-Frontends
An architecture that composes several independently built and deployed frontend bundles instead of one monolithic app — each owning a domain and shipping on its own cadence.
Multi-Agent System
An architecture where multiple specialized AI agents collaborate to handle complex workflows — each agent owns a domain and they hand off work to each other.
OAuth 2.0
The authorization framework (RFC 6749) that lets an application act on a user’s behalf with a scoped access token instead of the user’s password.
Outcome-Based Automation
AI automation defined by desired business results rather than prescribed step-by-step workflows, letting agents choose their own path.
PKCE (Proof Key for Code Exchange)
An OAuth 2.0 extension (RFC 7636) that binds an authorization code to the client that requested it, so an intercepted code cannot be exchanged for tokens.
Prompt Builder
Salesforce's tool for creating reusable, templatized prompts that pull in live CRM data. Build once, use across flows, agents, and record pages.
Prompt Chaining
Linking sequential LLM prompts so each output becomes the structured input for the next reasoning step.
Prompt Engineering
The practice of crafting precise instructions to guide an AI model's behavior, capabilities, and limitations.
Prompt Injection
An attack where malicious text hidden in data an AI reads hijacks its instructions, making it ignore its real task and do the attacker's bidding.
RAG (Retrieval-Augmented Generation)
A technique that makes AI smarter by fetching relevant information from your data before generating a response. The AI "looks it up" instead of guessing.
React External App (Salesforce)
A Multi-Framework React app for customers or partners, deployed with target Experience and served through Digital Experiences.
React Internal App (Salesforce)
A Multi-Framework React app for employees, deployed with target CustomApplication and opened from the App Launcher or the Salesforce mobile app.
Reasoning Model
A class of large language model trained to spend hidden internal "thinking" tokens before producing a user-facing answer — often dramatically improving performance on math, code, science, and complex multi-step problems compared to non-reasoning models of similar size.
Reverse Prompting
Instead of you prompting the AI, you let the AI prompt you. It interviews you for the context it needs first — then delivers a far better answer.
Salesforce
Salesforce is a leading cloud-based customer relationship management (CRM) platform that helps businesses manage their sales, customer service, and marketing efforts.
Salesforce Foundations
A free ($0) add-on for Salesforce Enterprise Edition and above that unlocks access to Agentforce, Data Cloud, and features from Sales, Service, Marketing, and Commerce clouds.
Salesforce Multi-Framework
A framework-agnostic Salesforce runtime that lets a standard React app run natively on the platform, with Salesforce authentication, security and governance applied automatically.
Salesforce Platform SDK
The client library (@salesforce/platform-sdk) a Multi-Framework React app uses to query and mutate Salesforce records over GraphQL, with no authentication or token management.
Service Cloud
Service Cloud is Salesforce's customer service platform that helps businesses manage and enhance their customer support operations.
Software Factory
A model for producing software through a standardized, repeatable, assembly-line-style pipeline rather than bespoke hand-crafting — so applications can be turned out predictably and at scale. In the agent era, the "workers" on the line are increasingly AI agents, and a spec or idea goes in one end while running, tested software comes out the other.
Test-Driven Development (TDD)
A software discipline where you write a failing test before writing the code that makes it pass — used for decades in traditional development and now extending to AI agents through "evals-as-tests."
Test-Time Compute
The principle that an AI model's output quality scales with the amount of compute it spends at inference time — not just with the size of the model. The architectural shift behind reasoning models like o1, Claude with extended thinking, and DeepSeek R1.
Token
The basic unit of text that AI models process — roughly 3/4 of a word. Token limits determine how much context an agent can use in a single interaction.
Token-Maxxing
The deliberate strategy of spending generous amounts of inference tokens — through extended thinking, deep research loops, or multi-shot agentic chains — to maximize output quality. The "more tokens equals better answers" doctrine that emerged with reasoning models. Also spelled "token-maxing."
Topics and Actions
The building blocks of Agentforce agents — topics define what an agent can discuss, actions define what it can do.
Trust Layer
Salesforce's AI trust and safety framework that ensures agents operate within compliance boundaries, prevent misuse, and maintain data security.
How to Use AIki
New to AI + Salesforce?
Start with these in order:
- 1. LLM - Understand the foundation
- 2. AI Agent - Understand the shift
- 3. Agentforce - Salesforce's implementation
- 4. Salesforce Foundations - Get started free
Coming soon: Einstein Copilot, Trust Layer, Apex, Flow, Experience Cloud, and more.
Have a term you want explained? Let us know
Ready to Put AI to Work?
Let's talk about what AI agents could do for your business. 30 minutes. No pitch deck. Just answers.
Book Intro Call