Create Agent Response

The Create Agent Response API endpoint allows generating a new response from a specific agent based on user input. This endpoint processes the user’s query, utilizes AI models, and optionally retrieves information from predefined sources before returning a response.

How This Endpoint Works

Requires the agent_id as a path parameter to specify which agent should generate the response.

Accepts a structured request body containing conversation history, AI model preferences, and additional configurations.

Returns an AI-generated response tailored to the provided context.

Request Body Details

History: A list of past messages exchanged between the user and the agent to maintain context.

Model: Specifies the AI model (e.g., gpt4-o) to be used for generating responses.

Chat Thread ID: Links the response to an ongoing conversation thread for better tracking.

Overrides: Allows enabling/disabling specific settings, such as retrieving responses from external sources.

Query Source: Defines the origin of the request (e.g., web)

Viewing Your Agent ID

If you are unsure where to find your agent_id, refer to this guide:
How to View Your Agent ID

curl --location --request POST '/v2/api/agent//response' \ --header 'Content-Type: application/json' \ --data-raw '{ "history": [ { "user": "hi", "bot": "Hello! How can I assist you today?" }, { "user": "Can you tell me about RAG from the sources?" } ], "model": "gpt4-o", "chat_thread_id": 4329, "overrides": { "sources": true }, "query_source": "web" }'

{ "success": true, "message": "Agent Response Created Successfully.", "data": { "answer": "Retrieval-Augmented Generation (RAG) is a technique in natural language processing that enhances the capabilities of generative language models by integrating real-time data retrieval. This approach addresses the limitations of traditional generative models, which rely solely on their training data and may produce outdated or incorrect information if the data is not current. RAG ensures that the responses generated are both contextually appropriate and grounded in factual information by incorporating real-time data from external sources.", "sources": "[0]", "followup_questions": [ "string" ], "indexes": [ "string" ], "query": "Explain how Retrieval-Augmented Generation (RAG) works.", "total_tokens": 4809, "prompt_tokens": 4329, "completion_tokens": 480, "cache_hit": false, "history": [ { "user": "string", "bot": "string" } ], "prompt_prefix": "You are an AI assistant specializing in providing legal advice on regulatory requirements, compliance issues, and legal barriers to market entry.", "instructions": [ "string" ], "agent_mode": "QnA", "references": [ { "number": 0, "url": "Introduction to RAG1738098993.txt", "order": 1 } ], "success": true, "corpus_ids": [ 645 ], "run_id": "bb7b49f8-fa06-4504-9e39-dec8f10cb0b9", "chat_thread_name": "RAG Process", "thread_id": 4354, "agent_response_id": 10954 } }

Request

Path Params

agent_id

integer

required

The ID of the agent for which the response is being created.

Body Params application/json

history

array [object {2}]

optional

The conversation history between the user and the bot.

user

string

optional

User's input in the conversation.

Example:

Can you tell me about RAG from the sources?

bot

string

optional

Bot's response to the user's input.

Example:

Sorry, I cannot find the answer in the available sources.

model

string

optional

The model to be used for generating the response.

Example:

gpt4-o

chat_thread_id

integer

optional

The ID of the chat thread.

Example:

4329

overrides

object

optional

Additional settings to override the response generation behavior.

sources

boolean

optional

Include sources in the response.

Example:

true

corpus_ids

array[integer]

optional

Narrow down multiple corpora to a few specific ones.

Example:

[645]

suggest_followup_questions

boolean

optional

Enable follow-up questions suggestion.

Example:

true

cache_skip

boolean

optional

Skip fetching the response from cache.

Example:

true

semantic_ranker

boolean

optional

Enable semantic ranking for the results.

Example:

true

semantic_captions

boolean

optional

Include semantic captions in the results.

Example:

false

top

integer

optional

Limit the number of results to this value.

Example:

query_source

string

optional

The source of the query.

Example:

web

Examples

Responses

🟢200OK

application/json

Response created successfully.

Body

success

boolean

optional

Whether the request was successful.

Example:

true

message

string

optional

Success message.

Example:

Agent Response Created Successfully.

data

object

optional

Response details.

answer

string

optional

The generated response for the user query.

Example:

Retrieval-Augmented Generation (RAG) is a technique in natural language processing that enhances the capabilities of generative language models by integrating real-time data retrieval. This approach addresses the limitations of traditional generative models, which rely solely on their training data and may produce outdated or incorrect information if the data is not current. RAG ensures that the responses generated are both contextually appropriate and grounded in factual information by incorporating real-time data from external sources.

sources

string

optional

Sources referenced during the response generation.

Example:

[0]

followup_questions

array[string]

optional

Suggested follow-up questions based on the response.

indexes

array[string]

optional

Indexes associated with the response.

query

string

optional

The user query that prompted the response.

Example:

Explain how Retrieval-Augmented Generation (RAG) works.

total_tokens

integer

optional

Total number of tokens used during the request.

Example:

4809

prompt_tokens

integer

optional

Number of tokens used for the prompt.

Example:

4329

completion_tokens

integer

optional

Number of tokens used for the completion.

Example:

480

cache_hit

boolean

optional

Indicates whether the response was fetched from the cache.

Example:

false

history

array [object {2}]

optional

Updated conversation history.

prompt_prefix

string

optional

Prefix used in the prompt.

Example:

You are an AI assistant specializing in providing legal advice on regulatory requirements, compliance issues, and legal barriers to market entry.

instructions

array[string]

optional

Additional instructions.

agent_mode

string

optional

The mode in which the agent operated.

Example:

QnA

references

array [object {3}]

optional

References used during the response generation.

success

boolean

optional

Indicates the success of the operation.

Example:

true

corpus_ids

array[integer]

optional

IDs of corpora used in response generation.

Example:

[645]

run_id

string

optional

ID of the specific execution run.

Example:

bb7b49f8-fa06-4504-9e39-dec8f10cb0b9

chat_thread_name

string

optional

Name of the chat thread.

Example:

RAG Process

thread_id

integer

optional

The ID of the chat thread.

Example:

4354

agent_response_id

integer

optional

The ID of the created agent response.

Example:

10954

Create Agent Response

How This Endpoint Works#

Request Body Details#

Viewing Your Agent ID#

Request

Responses

How This Endpoint Works

Request Body Details

Viewing Your Agent ID