Menu
Amazon Lex
Developer Guide

PostText

Sends user input (text-only) to Amazon Lex. Client applications can use this API to send requests to Amazon Lex at runtime. Amazon Lex then interprets the user input using the machine learning model it built for the bot.

In response, Amazon Lex returns the next message to convey to the user an optional responseCard to display. Consider the following example messages:

  • For a user input "I would like a pizza", Amazon Lex might return a response with a message eliciting slot data (for example, PizzaSize): "What size pizza would you like?"

  • After the user provides all of the pizza order information, Amazon Lex might return a response with a message to obtain user confirmation "Proceed with the pizza order?".

  • After the user replies to a confirmation prompt with a "yes", Amazon Lex might return a conclusion statement: "Thank you, your cheese pizza has been ordered.".

Not all Amazon Lex messages require a user response. For example, a conclusion statement does not require a response. Some messages require only a "yes" or "no" user response. In addition to the message, Amazon Lex provides additional context about the message in the response that you might use to enhance client behavior, for example, to display the appropriate client user interface. These are the slotToElicit, dialogState, intentName, and slots fields in the response. Consider the following examples:

  • If the message is to elicit slot data, Amazon Lex returns the following context information:

    • dialogState set to ElicitSlot

    • intentName set to the intent name in the current context

    • slotToElicit set to the slot name for which the message is eliciting information

    • slots set to a map of slots, configured for the intent, with currently known values

  • If the message is a confirmation prompt, the dialogState is set to ConfirmIntent and SlotToElicit is set to null.

  • If the message is a clarification prompt (configured for the intent) that indicates that user intent is not understood, the dialogState is set to ElicitIntent and slotToElicit is set to null.

In addition, Amazon Lex also returns your application-specific sessionAttributes. For more information, see Managing Conversation Context.

Request Syntax

Copy
POST /bot/botName/alias/botAlias/user/userId/text HTTP/1.1 Content-type: application/json { "inputText": "string", "sessionAttributes": { "string" : "string" } }

URI Request Parameters

The request requires the following URI parameters.

botAlias

The alias of the Amazon Lex bot.

botName

The name of the Amazon Lex bot.

userId

The ID of the client application user. Amazon Lex uses this to identify a user's conversation with your bot. At runtime, each request must contain the userID field.

To decide the user ID to use for your application, consider the following factors.

  • The userID field must not contain any personally identifiable information of the user, for example, name, personal identification numbers, or other end user personal information.

  • If you want a user to start a conversation on one device and continue on another device, use a user-specific identifier.

  • If you want the same user to be able to have two independent conversations on two different devices, choose a device-specific identifier.

  • A user can't have two independent conversations with two different versions of the same bot. For example, a user can't have a conversation with the PROD and BETA versions of the same bot. If you anticipate that a user will need to have conversation with two different versions, for example, while testing, include the bot alias in the user ID to separate the two conversations.

Length Constraints: Minimum length of 2. Maximum length of 100.

Pattern: [0-9a-zA-Z._:-]+

Request Body

The request accepts the following data in JSON format.

inputText

The text that the user entered (Amazon Lex interprets this text).

Type: String

Length Constraints: Minimum length of 1. Maximum length of 1024.

Required: Yes

sessionAttributes

Application-specific information passed between Amazon Lex and a client application. The value must be a JSON serialized and base64 encoded map with string keys and values.

For more information, see Setting Session Attributes.

Type: String to string map

Required: No

Response Syntax

Copy
HTTP/1.1 200 Content-type: application/json { "dialogState": "string", "intentName": "string", "message": "string", "responseCard": { "contentType": "string", "genericAttachments": [ { "attachmentLinkUrl": "string", "buttons": [ { "text": "string", "value": "string" } ], "imageUrl": "string", "subTitle": "string", "title": "string" } ], "version": "string" }, "sessionAttributes": { "string" : "string" }, "slots": { "string" : "string" }, "slotToElicit": "string" }

Response Elements

If the action is successful, the service sends back an HTTP 200 response.

The following data is returned in JSON format by the service.

dialogState

Identifies the current state of the user interaction. Amazon Lex returns one of the following values as dialogState. The client can optionally use this information to customize the user interface.

  • ElicitIntent – Amazon Lex wants to elicit user intent.

    For example, a user might utter an intent ("I want to order a pizza"). If Amazon Lex cannot infer the user intent from this utterance, it will return this dialogState.

  • ConfirmIntent – Amazon Lex is expecting a "yes" or "no" response.

    For example, Amazon Lex wants user confirmation before fulfilling an intent.

    Instead of a simple "yes" or "no," a user might respond with additional information. For example, "yes, but make it thick crust pizza" or "no, I want to order a drink". Amazon Lex can process such additional information (in these examples, update the crust type slot value, or change intent from OrderPizza to OrderDrink).

  • ElicitSlot – Amazon Lex is expecting a slot value for the current intent.

    For example, suppose that in the response Amazon Lex sends this message: "What size pizza would you like?". A user might reply with the slot value (e.g., "medium"). The user might also provide additional information in the response (e.g., "medium thick crust pizza"). Amazon Lex can process such additional information appropriately.

  • Fulfilled – Conveys that the Lambda function configured for the intent has successfully fulfilled the intent.

  • ReadyForFulfillment – Conveys that the client has to fulfill the intent.

  • Failed – Conveys that the conversation with the user failed.

    This can happen for various reasons including that the user did not provide an appropriate response to prompts from the service (you can configure how many times Amazon Lex can prompt a user for specific information), or the Lambda function failed to fulfill the intent.

Type: String

Valid Values: ElicitIntent | ConfirmIntent | ElicitSlot | Fulfilled | ReadyForFulfillment | Failed

intentName

The current user intent that Amazon Lex is aware of.

Type: String

message

A message to convey to the user. It can come from the bot's configuration or a code hook (Lambda function). If the current intent is not configured with a code hook or the code hook returned Delegate as the dialogAction.type in its response, then Amazon Lex decides the next course of action and selects an appropriate message from the bot configuration based on the current user interaction context. For example, if Amazon Lex is not able to understand the user input, it uses a clarification prompt message (for more information, see the Error Handling section in the Amazon Lex console). Another example: if the intent requires confirmation before fulfillment, then Amazon Lex uses the confirmation prompt message in the intent configuration. If the code hook returns a message, Amazon Lex passes it as-is in its response to the client.

Type: String

Length Constraints: Minimum length of 1. Maximum length of 1024.

responseCard

Represents the options that the user has to respond to the current prompt. Response Card can come from the bot configuration (in the Amazon Lex console, choose the settings button next to a slot) or from a code hook (Lambda function).

Type: ResponseCard object

sessionAttributes

A map of key-value pairs representing the session-specific context information.

Type: String to string map

slots

The intent slots (name/value pairs) that Amazon Lex detected so far from the user input in the conversation.

Type: String to string map

slotToElicit

If the dialogState value is ElicitSlot, returns the name of the slot for which Amazon Lex is eliciting a value.

Type: String

Errors

BadGatewayException

Either the Amazon Lex bot is still building, or one of the dependent services (Amazon Polly, AWS Lambda) failed with an internal service error.

HTTP Status Code: 502

BadRequestException

Request validation failed, there is no usable message in the context, or the bot build failed.

HTTP Status Code: 400

ConflictException

Two clients are using the same AWS account, Amazon Lex bot, and user ID.

HTTP Status Code: 409

DependencyFailedException

One of the downstream dependencies, such as AWS Lambda or Amazon Polly, threw an exception. For example, if Amazon Lex does not have sufficient permissions to call a Lambda function, it results in Lambda throwing an exception.

HTTP Status Code: 424

InternalFailureException

Internal service error. Retry the call.

HTTP Status Code: 500

LimitExceededException

Exceeded a limit.

HTTP Status Code: 429

LoopDetectedException

Lambda fulfilment function returned DelegateDialogAction to Amazon Lex without changing any slot values.

HTTP Status Code: 508

NotFoundException

The resource (such as the Amazon Lex bot or an alias) that is referred to is not found.

HTTP Status Code: 404

See Also

For more information about using this API in one of the language-specific AWS SDKs, see the following: