How to Ask

Humata allows you to chat with your document via the Ask endpoint. Create a conversation and then ask your documents

Ask in a Conversation

post

Ask a question in a conversation. Response is streamed via server-sent events.

Authorizations

ApiKeyAuth

AuthorizationstringRequired

API key provided as a Bearer token in the Authorization header.

Body

application/json

conversationIdstring · uuidRequired

modelstring · enumRequiredPossible values:

questionstringRequired

selectedAnswerApproachstring · enumOptionalPossible values:

Responses

200

Successfully answered the question.

text/event-stream

Responsestring

401

Unauthorized access, invalid API key.

application/json

429

Rate limit exceeded.

application/json

post

/v1/ask

HTTPcURLJavaScriptPython

POST /api/v1/ask HTTP/1.1
Host: app.humata.ai
Authorization: Bearer YOUR_SECRET_TOKEN
Content-Type: application/json
Accept: */*
Content-Length: 126

{
  "conversationId": "7290fd6b-fe15-4fd3-865e-8010bd8e8a38",
  "model": "gpt-4-turbo-preview",
  "question": "Who is George Washington?"
}

data: { 'content': 'George Washington was the first ' }
data: { 'content': 'president of the United States[3].\n\nReferences - [3]: Page 10 of History of the United States.pdf' }

Server-Sent Events

The Ask Endpoint returns a stream of server-sent events. Here is a full example of response to the question "Who is George Washington?":

data: { 'content': 'George Washington was ' }
data: { 'content': 'the first ' }
data: { 'content': 'president of the United States[3].' }
data: { 'content': '\n\nReferences - [3]: Page 10 of History of the United States.pdf' }

References

Each answer from Humata will provide references to the parts of the document that our AI used to answer the question. The format of references is subject to change.