Custom properties

Custom properties allows you to add any additional information to your logs, which can help you tag and filter logs.

What is a custom property?

A custom property is a key-value pair that you can add to an LLM log when sending an LLM log to Respan. It can help you tag and filter LLM logs.

Why use custom properties?

• Tag and filter LLM logs in your way.
• Add any additional information to your logs.

How to pass custom properties

You can pass custom properties to Respan by adding a metadata field to the request body.

OpenAI Python SDK

Here is an example of how to send 2 custom metadata to Respan in the OpenAI Python SDK.

from openai import OpenAI

client = OpenAI(
    base_url="https://api.respan.ai/api/",
    api_key="YOUR_RESPAN_API_KEY",
)

response = client.chat.completions.create(
    model="gpt-4o-mini",
    messages=[
        {"role": "user", "content": "Tell me a long story"}
    ],
    extra_body={
        "metadata": {
            "env": "production",
            "language": "en"
        }
    }
)

OpenAI TypeScript SDK

Here is an example of how to send 2 custom metadata to Respan in the OpenAI TypeScript SDK. In OpenAI TypeScript SDK, you should add a // @ts-expect-error before the metadata field.

import { OpenAI } from "openai";

const client = new OpenAI({
  baseURL: "https://api.respan.ai/api",
  apiKey: "YOUR_RESPAN_API_KEY",
});

const response = await client.chat.completions
  .create({
    messages: [{ role: "user", content: "Say this is a test" }],
    model: "gpt-4o-mini",
    // @ts-expect-error
    metadata: {
        "env": "production",
        "language": "en"
    }
  })
  .asResponse();

console.log(await response.json());

Standard API

Async logging

import requests

url = "https://api.respan.ai/api/request-logs/create/"
payload = {
    "model": "gpt-4o-mini",
    "prompt_messages": [
        {
            "role": "user",
            "content": "Hi"
        },
    ],
    "completion_message": {
        "role": "assistant",
        "content": "Hi, how can I assist you today?"
    },
    "metadata": {
        "env": "production",
        "language": "en"
    }
    # other parameters
}
headers = {
    "Authorization": "Bearer YOUR_RESPAN_API_KEY",
    "Content-Type": "application/json"
}

response = requests.request("POST", url, headers=headers, json=payload)

Other SDKs

We also support adding credentials in other SDKs or languages, please check out our integration section for more information.

Filter dashboard metrics

To use custom properties for filtering in the dashboard, you need to index the custom properties in the platform, otherwise the custom properties will not be available for filtering.

Filtering by custom properties in the dashboard is only available for Team plan and Enterprise plan users.

Filters in dashboard

After indexing the custom properties, you can filter the dashboard metrics by the custom properties.

Filter Logs

You can also filter logs by custom properties on the Logs page. Just click on the Filter button to filter the logs by the Custom properties.

Update custom properties

You can also update the custom properties of a log through the API.

Here are the steps to update the custom properties of a log:

1. Get unique_id of the log

   You need first to get the unique_id of the log you want to update. You can get the unique_id from the Logs list endpoint. OR you can get the unique_id from Logs page’s Log ID column.

2. Send the API request

   Pass the unique_id to the /api/request-logs/batch-update/ endpoint with the metadata and notes you want to update.

   Python

   import requests
   
   url = "https://api.respan.ai/api/request-logs/batch-update/"
   api_key = "YOUR_KEY" # Replace with your actual Respan API key
   data = {
       "logs": [
           {   "unique_id": "xxxxxx",
               "metadata": {
                   "env": "production",
                   "language": "en"
               },
               "note": "updated note"
           }
           # other logs to update
       ]
   }
   headers = {
       "Authorization": f"Bearer {api_key}",
       "Content-Type": "application/json"
   }
   
   response = requests.patch(url, headers=headers, json=data)
   print(response.json())