Using useTamboSuggestions

Overview

The useTamboSuggestions hook generates contextual prompt suggestions based on the conversation and available components. These suggestions help users discover what they can do next and keep conversations flowing naturally.

How It Works

When the AI sends a message, Tambo analyzes the conversation context
Suggestions are generated based on available components, tools, and conversation history
Users click a suggestion to populate the input field or auto-submit
The AI responds with the suggested action

Basic Setup

Import and use the hook in your chat component:

import { useTamboSuggestions, useTambo } from '@tambo-ai/react';

function Chat() {
  const { messages } = useTambo();
  const { suggestions, accept, isLoading } = useTamboSuggestions();

  if (isLoading) {
    return <div>Loading suggestions...</div>;
  }

  return (
    <div>
      {messages.map((msg) => (
        <Message key={msg.id} message={msg} />
      ))}
      
      {suggestions.length > 0 && (
        <div className="suggestions">
          {suggestions.map((suggestion) => (
            <button
              key={suggestion.id}
              onClick={() => accept({ suggestion })}
            >
              {suggestion.title}
            </button>
          ))}
        </div>
      )}
    </div>
  );
}

Auto-Submit Suggestions

Set shouldSubmit: true to automatically send the suggestion:

const { suggestions, accept } = useTamboSuggestions();

suggestions.map((suggestion) => (
  <button
    key={suggestion.id}
    onClick={() => accept({ suggestion, shouldSubmit: true })}
  >
    {suggestion.title}
  </button>
));

Custom Suggestion Count

Control how many suggestions to generate:

const { suggestions } = useTamboSuggestions({
  maxSuggestions: 5, // Generate up to 5 suggestions (1-10)
});

Configuration Options

maxSuggestions

Maximum number of suggestions to generate (1-10):

useTamboSuggestions({ maxSuggestions: 3 }); // Default: 3

autoGenerate

Whether to automatically generate suggestions after assistant messages:

useTamboSuggestions({ autoGenerate: false }); // Default: true

When false, call generate() manually:

const { generate, isGenerating } = useTamboSuggestions({ autoGenerate: false });

<button onClick={generate} disabled={isGenerating}>
  Generate Suggestions
</button>

queryOptions

Customize React Query behavior:

useTamboSuggestions({
  queryOptions: {
    staleTime: 5 * 60 * 1000, // Cache for 5 minutes
    retry: 3,
  },
});

Return Values

Data

const {
  suggestions,      // Array of suggestion objects
  data,            // Raw API response
  isLoading,       // Initial load state
  isSuccess,       // Successfully loaded
  isError,         // Error state
  error,           // Error object
  isFetching,      // Currently fetching (includes background)
} = useTamboSuggestions();

Actions

const {
  generate,        // Manually generate suggestions
  isGenerating,    // Generation in progress
  generateError,   // Generation error
  
  accept,          // Accept a suggestion
  isAccepting,     // Accepting in progress
  acceptError,     // Accept error
} = useTamboSuggestions();

UI State

const {
  selectedSuggestionId, // ID of last accepted suggestion
} = useTamboSuggestions();

Suggestion Object

Each suggestion has this structure:

interface Suggestion {
  id: string;                    // Unique identifier
  title: string;                 // Short title (display this)
  detailedSuggestion: string;    // Full prompt text (sent to AI)
}

Example:

{
  "id": "sugg_abc123",
  "title": "Show sales by region",
  "detailedSuggestion": "Create a bar chart showing sales data grouped by region for the last quarter"
}

Advanced Examples

Styled Suggestion Chips

import { useTamboSuggestions } from '@tambo-ai/react';
import { Sparkles } from 'lucide-react';

function SuggestionChips() {
  const { suggestions, accept, isAccepting, selectedSuggestionId } = useTamboSuggestions();

  if (suggestions.length === 0) return null;

  return (
    <div className="flex gap-2 flex-wrap mb-4">
      {suggestions.map((suggestion) => (
        <button
          key={suggestion.id}
          onClick={() => accept({ suggestion, shouldSubmit: true })}
          disabled={isAccepting}
          className={`
            flex items-center gap-2 px-4 py-2 rounded-full
            border border-gray-300 hover:border-blue-500
            transition-colors duration-200
            ${selectedSuggestionId === suggestion.id ? 'bg-blue-50 border-blue-500' : ''}
          `}
        >
          <Sparkles className="w-4 h-4" />
          <span className="text-sm">{suggestion.title}</span>
        </button>
      ))}
    </div>
  );
}

Manual Generation with Loading State

function SuggestionsPanel() {
  const { suggestions, generate, isGenerating, error } = useTamboSuggestions({
    autoGenerate: false,
  });

  return (
    <div>
      <button
        onClick={generate}
        disabled={isGenerating}
        className="mb-4 px-4 py-2 bg-blue-500 text-white rounded"
      >
        {isGenerating ? 'Generating...' : 'Get Suggestions'}
      </button>
      
      {error && (
        <div className="text-red-500 mb-4">
          Failed to generate suggestions: {error.message}
        </div>
      )}
      
      {suggestions.length > 0 && (
        <div className="grid gap-2">
          {suggestions.map((suggestion) => (
            <SuggestionCard key={suggestion.id} suggestion={suggestion} />
          ))}
        </div>
      )}
    </div>
  );
}

Suggestion Cards with Preview

function SuggestionCard({ suggestion }: { suggestion: Suggestion }) {
  const { accept } = useTamboSuggestions();
  const [showPreview, setShowPreview] = useState(false);

  return (
    <div className="border rounded-lg p-4 hover:shadow-lg transition-shadow">
      <div className="flex items-start justify-between">
        <div>
          <h3 className="font-medium">{suggestion.title}</h3>
          {showPreview && (
            <p className="text-sm text-gray-600 mt-2">
              {suggestion.detailedSuggestion}
            </p>
          )}
        </div>
        
        <div className="flex gap-2">
          <button
            onClick={() => setShowPreview(!showPreview)}
            className="text-sm text-gray-500 hover:text-gray-700"
          >
            {showPreview ? 'Hide' : 'Preview'}
          </button>
          
          <button
            onClick={() => accept({ suggestion, shouldSubmit: true })}
            className="px-3 py-1 bg-blue-500 text-white rounded text-sm"
          >
            Try it
          </button>
        </div>
      </div>
    </div>
  );
}

Conditional Display

Only show suggestions when appropriate:

function Chat() {
  const { messages, isStreaming } = useTambo();
  const { suggestions } = useTamboSuggestions();
  
  const latestMessage = messages[messages.length - 1];
  const shouldShowSuggestions = 
    !isStreaming &&                              // Not currently streaming
    latestMessage?.role === 'assistant' &&       // Last message from AI
    suggestions.length > 0;                      // Have suggestions

  return (
    <div>
      {messages.map((msg) => <Message key={msg.id} message={msg} />)}
      
      {shouldShowSuggestions && (
        <SuggestionChips suggestions={suggestions} />
      )}
    </div>
  );
}

Best Practices

Show suggestions after AI responses

Suggestions are most useful after the AI has responded, giving users natural next steps.

Hide during streaming

Don’t show suggestions while the AI is still responding:

const { isStreaming } = useTambo();
const { suggestions } = useTamboSuggestions();

if (isStreaming) return null;

return <SuggestionChips suggestions={suggestions} />;

Use clear, action-oriented titles

Suggestion titles should clearly indicate what will happen:

Good: “Show sales by region”, “Add a task”, “Export to CSV”
Bad: “Click here”, “Next”, “More”

Provide visual feedback

Show when a suggestion is being processed:

const { isAccepting, selectedSuggestionId } = useTamboSuggestions();

<button
  disabled={isAccepting}
  className={selectedSuggestionId === suggestion.id ? 'opacity-50' : ''}
>
  {suggestion.title}
</button>

Consider mobile layouts

Use horizontal scrolling or wrapping for mobile:

<div className="flex gap-2 overflow-x-auto pb-2 md:flex-wrap">
  {suggestions.map((s) => <SuggestionChip key={s.id} suggestion={s} />)}
</div>

Troubleshooting

No suggestions appearing

Ensure you’re using Tambo v1 API (not v0)
Check that messages exist and latest is from assistant
Verify autoGenerate is true (default)
Check browser console for errors
Ensure you have components registered

Suggestions not relevant

Register more components to give the AI more options
Add context helpers to provide app state
Increase maxSuggestions to get more variety

Suggestions taking too long

Reduce maxSuggestions (fewer = faster)
Check your network connection
Verify API endpoint is accessible

Next Steps

Voice Input

Add voice input to your chat

Thread Input Hook

Handle user input and submission

Component State

Manage state in AI components

Context Helpers

Improve suggestions with context

Get Started

Core Concepts

Guides

Templates & Examples

Using useTamboSuggestions

Overview

How It Works

Configuration Options

maxSuggestions

autoGenerate

queryOptions

Return Values

Data

Actions

UI State

Suggestion Object

Advanced Examples

Styled Suggestion Chips

Manual Generation with Loading State

Suggestion Cards with Preview

Conditional Display

Best Practices

Troubleshooting

Next Steps

Voice Input

Thread Input Hook

Component State

Context Helpers

Build docs developers (and LLMs) love

Get Started

Core Concepts

Guides

Templates & Examples

Documentation Index

​Overview

​How It Works

​Configuration Options

​maxSuggestions

​autoGenerate

​queryOptions

​Return Values

​Data

​Actions

​UI State

​Suggestion Object

​Advanced Examples

​Styled Suggestion Chips

​Manual Generation with Loading State

​Suggestion Cards with Preview

​Conditional Display

​Best Practices

​Troubleshooting

​Next Steps

Voice Input

Thread Input Hook

Component State

Context Helpers

Build docs developers (and LLMs) love

Overview

How It Works

Configuration Options

maxSuggestions

autoGenerate

queryOptions

Return Values

Data

Actions

UI State

Suggestion Object

Advanced Examples

Styled Suggestion Chips

Manual Generation with Loading State

Suggestion Cards with Preview

Conditional Display

Best Practices

Troubleshooting

Next Steps