Skip to content

Create an Agent (Step-by-step)

This guide walks you through creating a new agent in RhinoAgents from scratch.

1. Decide the Agent's Purpose

Start with a clear objective: lead qualification, support triage, invoice reconciliation, etc.

2. Create the Agent

  • Go to Agents → Create new agent.
  • Enter a name, description, and tag (for grouping).

3. Connect Data Sources

  • Attach integrations (CRM, sheets, database).
  • Upload credentials or add them via the Integrations page.

4. Design the Workflow

  • Use the visual builder to add steps: input, condition, action, API call, human review.
  • Add error-handling branches and timeouts.

5. Add Permissions & Roles

  • Assign which users can view, edit, or run the agent.
  • Add secrets as environment variables for the agent only.

6. Test the Agent

  • Use the built-in simulator to run sample inputs.
  • Inspect logs and step-by-step execution.

7. Activate & Monitor

  • Publish the agent to run on schedule, event triggers, or API calls.
  • Monitor metrics and add alerting (e.g., Slack, email) for failures.

Example YAML export (agent blueprint)

```yaml name: "InvoiceBot" description: "Automates invoice processing and reconciliation" triggers: - type: webhook path: /webhook/invoice workflow: - id: parse type: nlp model: invoice-parser-v1 - id: validate type: action service: accounting_api operation: validate_invoice - id: approve type: human_review role: finance_manager

docs/guides/create-chatbot.md

```markdown

Create a Chatbot (Step-by-step)

This guide covers building a user-facing chatbot using RhinoAgents.

1. Define the Use Case

Is this a support chatbot, sales assistant, or FAQ bot? Keep the scope narrow for the first iteration.

2. Create a New Agent (Chatbot)

  • Agents → Create → select Chatbot template (or blank).
  • Name it and set default language.

3. Author Conversation Flows

  • Build intents and dialog nodes in the visual UI or import a conversation JSON.
  • Add sample utterances and expected parameters (entities).

4. Add Integrations for Context

  • Connect CRM or database to fetch user data during conversations.
  • Add an authentication step if you need personalized data.

5. Add Fallbacks and Handoff

  • Design a fallback intent for unrecognized queries and escalate to a human when confidence is low.
  • Add a handoff action that creates a support ticket or opens a chat with an agent.

6. Test Conversationally

  • Use the chat preview to simulate user messages.
  • Validate NLU intent accuracy and entity extraction.

7. Deploy to Channels

  • Configure channel connectors: Webchat snippet, Slack, MS Teams, WhatsApp (via providers), or custom embedding.

Webchat snippet (example)

```html