The AI industry has a terminology problem. Every platform uses different words to describe similar concepts. One company says "agent." Another says "assistant." Others talk about workflows, copilots, skills, automations, tools, or functions.
At Rival, we believe AI should be easy to understand. Here's a simple guide to the core concepts you'll encounter across the Rival platform.
Agent: An AI system that can make decisions and take actions on its own.
Agents evaluate information, determine next steps, and execute tasks to achieve a goal.
Think intern. But one that actually follows directions.
Agentic AI: The type of AI that can understand a goal, make a plan, and take steps to get something done.
Unlike traditional AI that only generates responses, Agentic AI is designed to execute work.
Less "here's an idea." More "it's already done."
Agentic AI Platform: A platform for building, deploying, and managing AI systems that complete specific tasks given to it by humans.
Agentic AI platforms help organizations automate tasks, workflows, and business processes.
Because your best people should not be stuck doing your busy work
AI Assistant: An AI helper that users chat with directly.
Assistants answer questions, provide recommendations, generate content, and help users complete tasks.
The easiest way to get AI working for you.
AI Governance: The policies, controls, and oversight used to manage AI systems responsibly.
Governance helps organizations ensure AI is secure, compliant, and aligned with business objectives.
The grown-up in the room.
AI Sprawl: The uncontrolled growth of AI tools, models, agents, and workflows across an organization.
AI sprawl often leads to increased costs, duplication of tools, security risks, and governance challenges.
When every department discovers AI and nobody tells IT.
Audit Trail: A record of actions, decisions, and executions performed by AI systems.
Audit trails improve transparency, accountability, and governance.
For when Legal inevitably asks, "Who did this?"
Automation: The use of technology to perform work with minimal human involvement.
Automation reduces manual effort and increases operational efficiency.
A fancy word for "no more busywork."
BYOK (Bring Your Own Key): The ability to connect and use your own AI model credentials and accounts.
BYOK gives organizations greater flexibility, control, and ownership of their AI infrastructure.
Your models. Your rules.
Compliance: Ensuring AI systems meet legal, regulatory, and organizational requirements.
Compliance requirements vary by industry and geography.
The reason your lawyers still have jobs.
Connector: A connection between Rival and another application, system, or data source.
Connectors allow agents, workflows, and automations to access information and take actions across the tools you already use, such as email, calendars, CRMs, cloud storage, databases, and business applications.
Because AI can't automate what it can't access.
Cost Optimization: Reducing AI expenses by improving how models, workflows, and resources are used.
The goal is to maximize outcomes while minimizing unnecessary spend.
Good AI is powerful. Great AI is profitable.
Deterministic Function: A reliable action that produces the same result every time.
Deterministic functions are commonly used when consistency and predictability are required.
Same input. Same outcome. Every time.
Execution: The act of performing work.
Execution occurs when an AI system moves beyond recommendations and takes actions.
The difference between talking and doing.
Expert Workflow: A pre-built workflow by experts designed around a proven business process.
Expert Workflows help organizations deploy best practices faster.
Because somebody already made all the mistakes for you.
Function: A single task that performs a specific action.
Examples include sending an email, calling an API, generating a summary, or updating a CRM record.
One job. No excuses.
Guardrails: Rules and boundaries that define what an AI agent can and cannot do.
Guardrails help ensure agents operate safely, predictably, and within approved limits. They control behaviors such as what data an agent can access, which actions it can take, when approvals are required, and what tasks are prohibited.
Because not every decision should be automated.
Human-in-the-Loop (HITL): A process where humans review, approve, or guide AI actions.
Human oversight helps improve accuracy, governance, and trust.
Trust the robots, but keep humans in control.
Marketplace: A place where AI tools, workflows, agents, and assets can be shared and reused.
The Marketplace accelerates adoption by helping users discover existing solutions.
The app store for getting work done.
Matchmaker: The Rival system that finds the best existing tools, workflows, agents, and assets.
Its goal is to help users reuse proven solutions instead of rebuilding them.
Match made in automation heaven.
Memory: Information an agent remembers across conversations.
Memory allows an agent to retain important context, preferences, instructions, and knowledge between interactions. This memory is shared across everyone who uses the agent, helping it deliver more consistent and informed responses over time.
Because starting from scratch is for suckers
Model Routing: The process of selecting the best AI model for a specific task.
Routing helps optimize performance, cost, speed, and accuracy.
Don't use a Ferrari to pick up groceries.
Multi-Agent System: Multiple agents working together to complete a process or solve a problem.
Each agent may have a specialized responsibility within a larger workflow.
Like a team. But everyone shows up prepared.
Orchestration: The coordination of multiple systems, agents, workflows, and functions to achieve a goal.
Orchestration ensures all parts of a process work together effectively.
Getting all the robots to play nicely together.
Publisher: A user or organization that creates and shares assets in the Marketplace.
Publishers contribute reusable solutions that others can adopt.
The person saving everyone else from reinventing the wheel.
Ritual: A fixed routine that always runs the same way.
Examples include daily news briefings, weekly reports, and recurring executive updates.
Routine work without the routine.
RivalBot: The primary AI assistant within Rival.
RivalBot helps users discover tools, build workflows, automate work, and navigate the platform.
AI on easy mode.
Run: Executing a workflow, agent, or automation.
Every time a workflow performs work, a run occurs.
One job, start to finish.
Run Balance: The available resources or credits used to complete tasks on the Rival platform.
Run Balance determines how much automation can be performed.
Your AI fuel tank.
Schedules: Recurring automated runs that execute an agent on a predefined schedule.
Schedules allow agents to run automatically without user intervention. They can perform recurring tasks such as generating reports, monitoring systems, checking for updates, or delivering routine briefings at specified intervals.
The best work happens while you're not thinking about it.
Skill: Something an AI system knows how to do.
Examples include writing, summarizing, translating, researching, or analyzing information.
The building blocks of expertise.
Specialized Assistant: An AI assistant built for a specific job, role, or industry.
Examples include legal assistants, marketing assistants, financial assistants, and customer support assistants.
Purpose-built beats one-size-fits-all.
Sub-Agent: An agent that another agent can call to perform a specialized task.
Sub-agents allow agents to delegate work to other agents with specific expertise. Each sub-agent operates independently with its own instructions, tools, credentials, and context, enabling more complex workflows while keeping responsibilities separated.
Your agents subject matter expert.
Super Agent: A more advanced AI system with deeper capabilities.
Super Agents combine multiple skills, workflows, functions, tools, and data sources to solve larger business problems.
Like an agent that drank too much coffee and became management.
Task: A single unit of work performed by an agent, workflow, or user.
Tasks are the individual actions that make up larger workflows.
The “do this” part of getting things done.
Template: A reusable starting point for creating something new.
Templates help users launch workflows and agents faster.
A head start with instructions.
Tools: Any reusable component within the platform.
Tools may include workflows, agents, functions, templates, assistants, or automations.
Why rebuild something Dave already built six months ago?
Workflow: A series of steps that automate work.
Workflows connect tasks, systems, and actions together to complete a business process.
The assembly line for getting work off your plate.
Workspace: A shared environment where teams collaborate on AI systems and automations.
Workspaces help organizations manage users, assets, workflows, and governance.
Where the work happens and the excuses don't.
The future of AI isn't about prompts. It's about systems that can execute work.
Understanding the difference between assistants, agents, workflows, functions, and rituals is the first step toward building AI that delivers real business outcomes.
Rival gives organization the cheat code for easy AI.