What is a Rival Agent
An agent is an AI worker that takes action toward a goal. You give it instructions, context, and a set of tools, and it carries out real work — not just answering a question, but doing the task.
A Rival Agent is a customizable AI assistant created within your organization. Each agent has its own name, icon, persona, and safety boundaries. Beyond conversation, agents can run tools, reference knowledge sources, act inside third-party systems through connectors, hand work to sub-agents, and run on a schedule through rituals.
What a Rival Agent can do
Section titled “What a Rival Agent can do”Every agent starts with a persona and guardrails, then grows only the capabilities its job needs:
- Tools — run functions from the marketplace or your organization’s own collection.
- Knowledge — reference uploaded materials (PDFs, documents, text) when responding.
- Connectors — operate inside third-party platforms via secure sign-in (Slack, Microsoft 365, Google Workspace, Salesforce, and more).
- Sub-agents — delegate parts of a job to other agents. See Sub-agents & Multi-agent systems.
- Rituals — run automatically on a schedule with no person in the chat. See Rituals.
An agent can stay purely conversational or grow into a full process-handling system. You configure only what your use case requires.
How an agent differs from a tool
Section titled “How an agent differs from a tool”A tool is a discrete function: it takes input, runs code, and returns output. An agent pairs a language model with a persona and decides which tools to call to handle a request. Agents talk conversationally; tools require explicit, structured calls. Agents use tools — tools can’t invoke agents — and an agent can delegate to other agents as sub-agents.
Configuration sections
Section titled “Configuration sections”An agent editor is organized into:
- Basic identity — name, description, instructions, and theme.
- Personality & Guardrails — communication style and permitted vs. prohibited actions. See Personality & Guardrails.
- Tools & Memory — attached tools, knowledge uploads, connections.
- Sub-agents — agents available for delegation.
- Rituals — autonomous scheduled runs.
- Connectors — third-party integrations.
What’s different in the enterprise
Section titled “What’s different in the enterprise”The mechanics above are the same across Rival. What enterprise adds is governance:
- Team members adopt from an approved set, not the whole marketplace — see Adopting approved agents.
- Publishing an agent for others routes through admin approval — see Publishing an Agent.
- Org-wide guardrails and allowed models set by an admin apply on top of each agent’s own — see Guardrails & allowed models.
Publishing visibility runs Private, Organizational, or Public, with versioning and release notes; public agents go through marketplace review.