Skip to main content
Through the Olostep /v1/answers endpoint you can search the web with natural language and return AI‑powered answers and data in the JSON shape you want. This lets you ground your products on real-world data and sources, enrich data points or spreadsheets
  • Ask a question or give AI a data point you want to enrich
  • Optional: specify the JSON structure you want back
It will:
  • Search, clean, validate and return the data it found on the Web
  • Return sources used to generate the answer
  • Handle uncertainty with NOT_FOUND values when data cannot be verified
For details, see the Answers Endpoint API Reference. By default we use a generic web index and a cost‑efficient LLM validator. Enterprise customers have access to proprietary industry specific web indexes, exclusive private data (including phone numbers and emails) and custom LLM models most suited for their use case. Contact us for access: info@olostep.com

​
Use cases

The answers endpoint can be used to:
  • Ground AI applications on real world data and facts
  • Enrich spreadsheets and data points for Recruting, Finance, Consulting, and Sales
Here’s a demo of an AI powered spreadsheet powered by the Answers endpoint: https://www.olostep.com/demos/spreadsheet-enrich

​
Installation

# pip install requests

import requests

​
Usage

Ask a question and pass a JSON schema to guide the output. You can also not pass the json parameter and the API will return a json object with the answer text inside the result field. 
import requests, json

endpoint = "https://api.olostep.com/v1/answers"
payload = {
  "task": "What is the latest book by J.K. Rowling?",
  "json": {"book_title": "", "author": "", "release_date": ""}
}
headers = {"Authorization": "Bearer <YOUR_API_KEY>", "Content-Type": "application/json"}

response = requests.post(endpoint, json=payload, headers=headers)
print(json.dumps(response.json(), indent=2))

​
Response

Like other Olostep endpoints, you will receive a answer object in response. The answer object has a few properties like id and result. 
{
  "id": "answer_9bi0sbj9xa",
  "object": "answer",
  "created": 1760327323,
  "metadata": {},
  "task": "What is the latest book by J.K. Rowling?",
  "result": {
    "json_content": "{\"book_title\":\"The Hallmarked Man\",\"author\":\"J.K. Rowling (as Robert Galbraith)\",\"release_date\":\"2 September 2025\"}",
    "json_hosted_url": "https://olostep-storage.s3.us-east-1.amazonaws.com/answer_9bi0sbj9xa.json",
    "sources": [
      "https://strikefans.com/the-books/",
      "https://www.facebook.com/groups/496943608606523/posts/1136830134617864/",
      "https://robert-galbraith.com/strike-books/",
      "https://www.novelsuspects.com/series-list/robert-galbraith-cormoran-strike-series-in-order/",
      "https://www.reddit.com/r/books/comments/1na833a/jk_rowlings_new_strike_novel_900_pages_of_romance/",
      "https://www.harrypotter.com/writing-by-jk-rowling",
      "https://stories.jkrowling.com/book-news/",
      "https://deadline.com/2024/09/jk-rowling-writing-futuristic-novel-1236093909/",
      "https://www.reddit.com/r/FantasticBeasts/comments/1cl1shn/jk_rowling_may_2024_ive_got_six_more_books_in_my/",
      "https://www.jkrowling.com/news/"
    ]
  }
}
Your requested answer, formatted according to the json parameter, is in response.result.json_content and the list of sources in response.result.sources. You can parse the stringified JSON to access the structured data. 
{
  "book_title": "The Hallmarked Man",
  "author": "J.K. Rowling",
  "release_date": "September 2, 2025"
}
Sources example: 
[
    "https://www.harrypotter.com/writing-by-jk-rowling",
    "https://stories.jkrowling.com/book-news/",
    "https://deadline.com/2024/09/jk-rowling-writing-futuristic-novel-1236093909/",
    "https://www.reddit.com/r/FantasticBeasts/comments/1cl1shn/jk_rowling_may_2024_ive_got_six_more_books_in_my/",
    "https://www.jkrowling.com/news/"
]
When you don’t pass the json parameter, the API will return a json object with the answer text inside the result field. 
{
  "result": "The latest book by J.K. Rowling is The Hallmarked Man."
}

​
Flexible json parameter

  • Provide a JSON object with empty values as a schema, or a string describing the data you want.
  • If the agent isn’t confident, it returns NOT_FOUND for that field.
Python
import requests, json

endpoint = "https://api.olostep.com/v1/answers"
payload = {
  "task": "how much did Olostep raise?", 
  "json": {
    "amount": ""
  }
}
headers = {
  "Authorization": "Bearer <YOUR_API_KEY>",
  "Content-Type": "application/json"
}
response = requests.post(endpoint, json=payload, headers=headers)
print(json.dumps(response.json(), indent=2))
This would return: 
{
  "amount": "NOT_FOUND"
}