Conversational

/chat

Published @ 17/11/23 by

Fatos, Founder @ answerly.io

POST https://ai.api.answerly.io/chat

Triggers the agent to generate a response based on the provided message.

Request Body

agentId string Required

The ID of your agent. You can read more about how to get the agent ID by clicking here.

conversationId string Required

The ID of your conversation. You can leave this field as an empty string, if you want to initiate a new conversation.

message string Required

The message that you want the agent to respond to.

channel string Required

The channel that you need the response for.

Possible properties: web

responseStyle string Required

The style that you want the response to be in.

Possible properties: plaintext markdown html

actionRequest string Required

actionRequest is a key that has been created for future use, and does not have any particular effect at this time.

Please use the same value for this key as shown on the example to the right.

Example

curl --request POST \
  --url https://ai.api.answerly.io/chat/ \
  --header 'Content-Type: application/json' \
  --data '{
	"agentId": "YOUR_AGENT_ID",
	"conversationId":"",
	"message": "YOUR_MESSAGE",
	"channel": "web",
	"responseStyle": "html",
	"actionRequest": {
		"name": "conversational"
	}
}'

Return (success)

status boolean

A boolean indicating the success or failure of the request.

data.conversationId string

Represents the ID for your conversation. If no conversation ID was provided, this field will display the newly assigned conversation ID.

data.messages string[]

An array containing responses from the agent.

These responses represent paragraphs, or collection of paragraphs, and collectively form the complete message sent by the agent.

data.actionResponse string[]

actionResponse is a key that has been created for future use, and does not have any particular effect at this time.

data.meta.tokens.conversational number

The amount of tokens spent to generate the agent response.

data.meta.tokens.functional number

The amount of tokens spent on AI features, such as human takeover detection.

data.meta.tokens.embeddings number

The amount of tokens spent on converting your message into a vector embedding.

data.meta.source.$.dataItemId number

The ID of the data item that helped the agent generate the response.

data.meta.source.$.cosineDistanceScore number

The coisine similarity score between your message, and the data item's content.

Example

{
    "status": true,
    "data": {
      "conversationId": "NEW_CONVERSATION_ID",
      "messages": [
        "AGENT_MESSAGE_PART_1",
        "AGENT_MESSAGE_PART_2"
      ],
      "actionResponse": {
        "name": "conversational"
      },
      "meta": {
        "tokens": {
          "conversational": 556,
          "functional": 143,
          "embeddings": 4
        },
        "source": [
          {
            "dataItemId": "e85e1a83-187b-4d2e-a160-7632af8465c1",
            "cosineDistanceScore": 0.18312443394210398
          }
        ]
      }
    }
  }

Return (failure)

status boolean

A boolean indicating the success or failure of the request.

data string

An error message returned by the server.

Example

{
    "status": false,
    "data": "An agent with that ID was not found"
  }