OrelloOrello

Orello Developer Center

Build, customize, and deploy conversational customer support AI agents. Ground them with your business files and URLs, and embed them natively in any front-end stack.

Quickstart

Add real-time voice and text conversational automation to your platform. In just 3 steps, you can embed your AI support widget grounded with custom knowledge.

Quickstart

Embed your first AI Agent in under 5 minutes.

SDKs & Libraries

Native frameworks (React, Next.js, Vue) and CDN script.

Authentication

Public and secret API keys, environments, and safety.

Assistants API

Programmatically configure models, tone, and settings.

Knowledge Base

Upload files and URLs to ground assistant replies.

Webhooks

Real-time conversation events and voice action triggers.

1Create an Assistant

Head over to the Orello Dashboard to configure your assistant tone (formal, casual, sales, support), select your LLM model provider, and generate your API tokens.

2Implementation

Install the client package and mount the agent widget at the root of your application.

$npm install @orello/react
App.tsx
import { OrelloAgent } from '@orello/react';

function App() {
  return (
    <div className="app-container">
      <YourMainApp />
      {/* Grounded customer support agent */}
      <OrelloAgent
        agent="your-agent-id"
        apiKey="pk_live_your_public_key"
      />
    </div>
  );
}

SDKs & Libraries

Whether you build in Next.js, React, Vue, Svelte, or plain static HTML, Orello provides modular native components that plug straight into your build pipeline.

$npm install @orello/react
App.tsx
import { OrelloAgent } from '@orello/react';

function App() {
  return (
    <div className="app-container">
      <YourMainApp />
      {/* Grounded customer support agent */}
      <OrelloAgent
        agent="your-agent-id"
        apiKey="pk_live_your_public_key"
      />
    </div>
  );
}

Authentication

The Orello API utilizes two types of credentials to authenticate requests, ensuring strict security boundaries depending on whether the code executes in a client browser or on a private server.

Public Client Key

Client Safe
pk_live_••••••••

Safe to expose in frontend components (React, HTML). Restricts operations strictly to initiating conversations and widget interactions.

Secret Admin Key

Server Only
sk_live_••••••••

Highly sensitive. Keep private and use ONLY on backends or environments like Node scripts. Enables full training, grounding, and billing access.

All server-side requests must include the secret admin key in the HTTP headers using Bearer token format:

Request Header
Authorization: Bearer sk_live_or_••••••••••••••••

Assistants API

The Assistants API allows you to programmatically create, manage, configure, and inspect AI support agents. Customize parameters such as AI providers, response length, voice voices, and prompt guidelines.

POST/v1/assistantCreate a new AI support assistant
GET/v1/assistant/:idRetrieve agent configs, tone, and active model
PATCH/v1/assistant/:idUpdate prompt boundaries, tone, speech parameters
POST/v1/assistant/:id/deployDeploy assistant settings to development or production

Programmatic Creation Example

Create an agent dynamically using your server-side Secret Key:

Node.js Client
const createAssistant = async () => {
  const res = await fetch("https://api.orello.space/api/v1/assistant", {
    method: "POST",
    headers: {
      "Authorization": "Bearer sk_live_your_secret_key",
      "Content-Type": "application/json"
    },
    body: JSON.stringify({
      name: "Product Support Expert",
      tone: "support", // formal | casual | sales | support
      responseStyle: "balanced", // short | balanced | detailed
      provider: "openai" // openai | google | anthropic
    })
  });
  const data = await res.json();
  console.log("Created Assistant UID:", data.data.uid);
};

Knowledge Base Grounding

Ground your support assistants in your actual docs, product features, and FAQs to prevent AI hallucinations. By grounding responses, the agent draws information solely from uploaded facts and corporate logic.

POST/v1/assistant/:id/knowledgeUpload and index custom training files (PDF, txt, csv) or web URL anchors
PATCH/v1/assistant/:id/knowledge/:knowledgeUidUpdate indexed metadata or replace reference details
DELETE/v1/assistant/:id/knowledge/:knowledgeUidRemove knowledge item, deleting vector search references

Grounding via cURL Upload

Upload business knowledge programmatically using a standard multipart form request:

cURL Request
curl -X POST https://api.orello.space/api/v1/assistant/asst_10243/knowledge \
  -H "Authorization: Bearer sk_live_your_secret_key" \
  -F "file=@/path/to/product_api_guide.pdf" \
  -F "type=document"

Webhooks & Events

Listen to real-time events triggered by Orello. Get notified instantly when an assistant starts a conversation, triggers custom action functions, or terminates a live support call.

Event Event NamePayload Triggers
conversation.startedUser triggers conversational widget or starts a call session
conversation.completedAI assistant completes task, resolves inquiry, or closes window
agent.action_triggeredAI assistant invokes a custom registered function tool (e.g. cancels subscription)
call.startedUser starts voice-to-voice communication using web speech portals
call.endedVoice communication session closes, triggering token billing and feedback calculations

Handling Action Webhooks

Receive webhook POST events on your Node.js or Next.js backend and execute custom business actions:

Express Node.js Route
app.post("/webhooks/orello", (req, res) => {
  const event = req.body;

  if (event.type === "agent.action_triggered") {
    const { toolName, arguments: args } = event.data;
    console.log(`Agent called tool ${toolName} with args:`, args);

    // Implement custom logic (e.g., query database or upgrade package)
  }

  res.status(200).send({ received: true });
});

Ready to automate customer support?

Create your first agent, index your web resources, and start resolving support tickets 24/7 today.

Create Agent FreeView GitHub SDKs