AIAnswer
creates an AI-driven answer UI component that can interact with a dataset to provide relevant answers based on user inputs. It leverages the power of machine learning to understand the user's questions and fetch the most relevant information.
Example uses:
- Providing answers to common questions related to a specific topic.
- Offering support and help by answering user queries.
- Building a knowledge base or FAQ section for your website.
Usage
Basic Usage
<AIAnswer
componentId="AIAnswerComponent"
/>
Usage With All Props
<AIAnswer
componentId="AIAnswerComponent"
title="AI Answer"
placeholder="Ask something..."
react={{
and: ['AIAnswer'],
}}
onData = {({
data,
rawData,
loading,
error
}) =>{
// perform side effects
}
renderError={(error, handleRetry) => <div><pre>{JSON.stringify(error)}</pre><button onClick={handleRetry}>Try Again</button></div>}
showVoiceInput={true}
showIcon={true}
icon={<span>🚀</span>}
iconPosition="left"
enterButton={true}
renderEnterButton={(onClickCb) => <button onClick={onClickCb}>Ask me!</button>}
AIConfig={{
docTemplate: '${source.text} is ${source.summary} with url as ${source.url}',
queryTemplate: 'Answer the following: ${value}',
topDocsForContext: 7,
maxTokens: 4000,
systemPrompt: "You're a helpful assistant."
}}
showFeedback={true}
showSourceDocuments={true}
renderSourceDocument={(obj) => {
return (
<span>❤️ {obj.original_title}
</span>
)}}
onSourceClick={(srcObj)=>{
// perform any side effects
}}
triggerOn="always" // or "question", "manual" (default is always)
renderTriggerMessage={<div>End your query with a question mark (?)</div>} // string or JSX
/>
Props
componentId
Type | Optional |
---|---|
String |
No |
unique identifier of the component, can be referenced in other components' react
prop.
title
Type | Optional |
---|---|
String or JSX |
Yes |
set the title of the component to be shown in the UI.
AIConfig
Type | Optional |
---|---|
Object |
Yes |
Specify additional options for configruing AI responses.
Accepts the following properties:
- systemPrompt
String
[optional] The system prompt to send as the first message to ChatGPT. Defaults toYou are a helpful assistant
. - docTemplate
String
[optional] Template to use for building the message sent to ChatGPT for every hit of the response. ThedocTemplate
string supports dynamic values using the special syntax${}
. These values are resolved while the ChatGPT request body is built. It supports keys that are present in the_source
field in the response hits. As an example,source.title
will resolve to_source.title
. If values are not found, defaults to an empty string. - queryTemplate
String
[optional] Template to use for building the message that is sent to ChatGPT as the final question. Defaults toCan you tell me about ${value}
wherevalue
is thequery.value
. The querytemplate string supports a dynamic value ofvalue
which is the query.value of the query. - topDocsForContext
number
[optional] Number of docs to use to build the context. Defaults to 3. This has an upper limit as the total number of hits returned. - maxTokens
number
[optional] The maximum tokens that can be used for the output. Deafults to being dynamically calculated. Accepts a value between [1, 8000]. - temperatue
number
[optional] A control for randomness, a lower value implies a more deterministic output. Defaults to 1, valid values are between [0, 2].
placeholder
Type | Optional |
---|---|
String |
Yes |
set placeholder text to be shown in the component's input field. Defaults to "Search".
showIcon
Type | Optional |
---|---|
Boolean |
Yes |
whether to display a search or custom icon in the input box. Defaults to true
.
iconPosition
Type | Optional |
---|---|
String |
Yes |
sets the position of the search icon. Can be set to either left
or right
. Defaults to right
.
icon
Type | Optional |
---|---|
JSX |
Yes |
set a custom search icon instead of the default 🔍 . Takes precedence over iconURL
prop.
iconURL
Type | Optional |
---|---|
string |
Yes |
Customize search icon by providing a URL.
showInput
Type | Optional |
---|---|
Boolean |
Yes |
show the input element to allow asking follow up questions. Defaults to true
.
enterButton
Type | Optional |
---|---|
boolean |
Yes |
When set to true
, the results would only be updated on press of the button. Defaults to false
. You can also provide styles using the ai-enter-button
key in the innerClass
prop.
<AIAnswer
id="ai-component"
enterButton={true}
/>
renderEnterButton
Type | Optional |
---|---|
Function |
Yes |
renders a custom jsx markup for the enter button. Use in conjunction with enterButton
prop set to true
.
<AIAnswer
id="ai-component"
enterButton
renderEnterButton={clickHandler => (
<div
style={{
height: '100%',
display: 'flex',
alignItems: 'stretch',
}}
>
<button style={{ border: '1px solid #c3c3c3' }} onClick={clickHandler}>
🔍 Search
</button>
</div>
)}
/>
showVoiceInput
Type | Optional |
---|---|
Boolean |
Yes |
show a voice icon in the AIAnswer to enable users to set voice input. Defaults to false
.
showFeedback
Type | Optional |
---|---|
Boolean |
Yes |
Toggles displaying the feedback UI component to record AI session's feedback. Defaults to true.
showSourceDocuments
Type | Optional |
---|---|
Boolean |
Yes |
Whether to show the documents from which the AIAnswer is generated or not. Defaults to true
.
renderSourceDocument
Type | Optional |
---|---|
Function |
Yes |
Render a custom label by returning string or JSx. Default label is rendered as the resolved value of _id
when showSourceDocument
is set to true.
<AIAnswer
id="ai-component"
showSourceDocuments={true}
renderSourceDocument={(obj) => {
return <span>❤️ {obj.original_title}</span>;
}
}
/>
onSourceClick
Type | Optional |
---|---|
Function |
Yes |
callback to handle side-effects when a source button is clicked. Accepts a sourceObj
param associated with the source button clicked.
<AIAnswer
id="ai-component"
showSourceDocuments={true}
renderSourceDocument={(obj) => {
return <span>❤️ {obj.original_title}</span>;
}
}
onSourceClick={(srcObj)=>{
// perform any side effects
}}
/>
triggerOn
Type | Optional |
---|---|
string |
Yes |
sets the trigger for fetching the AI answer. Defaults to always
, Accepts one of the following three:
- manual
- AI answer is shown when triggered using the askButton
or clicking the first suggestion item(renderTriggerMessage
).
- question
- AI answer is shown when the user searches for a question.
- always
- AI answer is always shown for every query.
<AIAnswer
id="ai-component"
triggerOn="manual"
/>
renderTriggerMessage
Type | Optional |
---|---|
JSX |
Yes |
custom markup to display as the first suggestion in the dropdown to trigger the AI answer. It should be used along with enableAI
set to true
.
<AIAnswer
id="ai-component"
triggerOn="manual"
renderTriggerMessage={<div>End your query with a question mark (?)</div>} // string or JSX
/>
render
Type | Optional |
---|---|
Function |
Yes |
You can render suggestions in a custom layout by using the render
prop.
It accepts an object with these properties:
loading
:boolean
indicates that the query is still in progress.error
:object
An object containing the error info.data
:array
An array of messages obtained from the conversation with AI service.rawData
object
An object of raw response as-is from elasticsearch query.
<AIAnswer
componentId="ai-answer"
render={({ loading, data, error, rawData }) => {
if (loading) {
return 'loading...';
}
if (error) {
return <pre>{JSON.stringify(error)}</pre>;
}
if (data && Array.isArray(data)) {
return (
<div style={{ width: '80%', margin: '0 auto', padding: '20px' }}>
{data.map((message, index) => {
const isSender = message.role === 'user';
const messageStyle = {
backgroundColor: isSender ? '#cce5ff' : '#f8f9fa',
padding: '10px',
borderRadius: '7px',
marginBottom: '10px',
maxWidth: '80%',
alignSelf: isSender ? 'flex-end' : 'flex-start',
display: 'inline-block',
border: '1px solid',
color: isSender ? '#004085' : '#383d41',
position: 'relative',
whiteSpace: 'pre-wrap',
};
return (
<div
key={index}
style={{
display: 'flex',
justifyContent: isSender
? 'flex-end'
: 'flex-start',
}}
>
<div style={messageStyle}>{message.content}</div>
</div>
);
})}
</div>
);
}
}}
/>
Or you can also use render function as children
<AIAnswer>
{
({
loading,
error,
data,
rawData,
}) => (
// return UI to be rendered
)
}
</AIAnswer>
renderError
Type | Optional |
---|---|
String or JSX or Function |
Yes |
can be used to render an error message in case of any error.
renderError={(error, handleRetry) => (
<div>
Something went wrong!<br/>Error details<br/>{error}
<button onClick={handleRetry}>Regenerate response</button>
</div>
)
}
getMicInstance
Type | Optional |
---|---|
Function |
Yes |
You can pass a callback function to get the instance of SpeechRecognition
object, which can be used to override the default configurations.
renderMic
Type | Optional |
---|---|
String or JSX or Function |
Yes |
can we used to render the custom mic option.
It accepts an object with the following properties:
handleClick
:function
needs to be called when the mic option is clicked.status
:string
is a constant which can have one of these values:
INACTIVE
- mic is in inactive state i.e not listening
STOPPED
- mic has been stopped by the user
ACTIVE
- mic is listening
DENIED
- permission is not allowed
renderMic = {({ handleClick, status }) => {
switch(status) {
case 'ACTIVE':
return <img src="/active_mic.png" onClick={handleClick} />
case 'DENIED':
case 'STOPPED':
return <img src="/mute_mic.png" onClick={handleClick} />
default:
return <img src="/inactive_mic.png" onClick={handleClick} />
}
}}
onData
Type | Optional |
---|---|
Function |
Yes |
You can pass a callback function to listen for the changes in suggestions. The function receives data
, rawData
, loading
and error
as a single parameter object.
<AIAnswer
componentId="AIAnswerComponent"
// ... other props ...
onData = {({
data,
rawData,
loading,
error
}) =>{
// do something with the updated properties
}}
/>
onError
Type | Optional |
---|---|
Function |
Yes |
You can pass a callback function that gets triggered in case of an error and provides the error
object which can be used for debugging or giving feedback to the user if needed.
Demo
Styles
AIAnswer
component supports an innerClass
prop to provide styles to the sub-components of AIAnswer. These are the supported keys:
ai-title
ai-search-mic
ai-enter-button
ai-error
ai-message
ai-input
ai-feedback
Read more about it here.