Upsonic integration

What is Upsonic

Upsonic is an open-source Python framework for building AI agents. It supports any LLM (e.g., OpenAI, Anthropic, Azure, and Bedrock) and provides:

An AutonomousAgent class with built-in shell and filesystem access
A safety engine for policy-based content filtering
Built-in OCR and document processing
Session and long-term memory
Multi-agent coordination and MCP server support
Universal Commerce Protocol (UCP) tools for AI-driven commerce workflows

See the Upsonic documentation for more details on building AI agents.

Help keep this page up to date

This integration uses a third-party service. If you find outdated content, please submit an issue on GitHub.

How to use Apify with Upsonic

This guide shows how to integrate Apify Actors with Upsonic using the built-in ApifyTools class. The example builds a restaurant discovery agent that uses the Google Places Crawler Actor to search Google Maps with a natural language query. It saves the results to a Markdown file.

Prerequisites

Apify API token: Obtain your API token from the Apify Console.
Anthropic API key (or another LLM provider): Get one from the Anthropic Console.
Python environment: Python 3.9 or later.
Required packages: Install the following dependencies:

pip install 'upsonic[custom-tools]' python-dotenv apify-client anthropic

Alternative LLM providers

Upsonic uses LiteLLM under the hood. Swap anthropic/claude-sonnet-4-6 for any supported model string, e.g. openai/gpt-4o or gemini-3-flash-preview. Update the environment variable and model name accordingly.

Basic integration example

Set up an Upsonic agent with Apify tools. This example uses the RAG Web Browser Actor to extract content from a URL.

import os
from upsonic import Agent, Task
from upsonic.tools.custom_tools.apify import ApifyTools

agent = Agent(
    "anthropic/claude-sonnet-4-6",
    tools=[
        ApifyTools(
            actors=["apify/rag-web-browser"],
            apify_api_token=os.getenv("APIFY_API_TOKEN"),
        )
    ],
)

task = Task("What information can you find on https://docs.apify.com/platform/actors?")
agent.print_do(task)

Running this code fetches and summarizes the content from the provided URL.

Advanced scenario: Restaurant discovery agent

Combine ApifyTools with the Google Places Crawler to build an agent that accepts a natural language query and returns a list of matching restaurants.

actor_defaults pre-sets configuration that never needs to change - such as result limits and output format - while leaving the search query under the LLM's control. The timeout override is necessary because the Actor takes 60-90 seconds to crawl Google Maps.

import os
from upsonic import Agent, Task
from upsonic.tools.custom_tools.apify import ApifyTools
from dotenv import load_dotenv

load_dotenv()

agent = Agent(
    "anthropic/claude-sonnet-4-6",
    tools=[
        ApifyTools(
            actors=["compass/crawler-google-places"],
            apify_api_token=os.getenv("APIFY_API_TOKEN"),
            actor_defaults={
                "compass/crawler-google-places": {
                    "maxCrawledPlacesPerSearch": 10,
                    "maxImages": 0,
                    "outputFormats": ["markdown"],
                }
            },
            timeout=180.0,
            max_retries=0,
        )
    ],
)

task = Task("Find cheap and tasty falafel places in Kadikoy, Istanbul")
agent.print_do(task)

with open("results.md", "w") as f:
    f.write(task.response)

print("Results saved to results.md")

The agent interprets the natural language query, calls the Actor with appropriate search parameters, and formats the response:

## Best cheap and tasty falafel places in Kadikoy

### 1. Falafella - 4.3/5
- Price: very affordable
- Address: Caferaga, Moda Cd. No:53A, Kadikoy
- Hours: 11 AM - 2 AM
- Vegan options available

### 2. Nohut Falafel & Humus - 4.8/5
- Address: Osmanaga, Sakiz Sk. No:22C, Kadikoy
- Hours: 12 PM - 10 PM
- Gluten-free and vegan, known for fresh ingredients

Crawl time

Each run takes 60-90 seconds as the Actor crawls Google Maps. Keep maxCrawledPlacesPerSearch at 10 or below - more results can exceed the model's context limit.

Configuration options

actors (string or List[string], default: None) : Single Actor ID or list of Actor IDs to register as tools.

apify_api_token (string, default: None) : Apify API token. Falls back to the APIFY_API_TOKEN environment variable.

actor_defaults (Dict[string, Dict[string, Any]], default: None) : Per-actor default input values. Keys are Actor IDs, values are dicts of parameter to value. Hidden from the LLM and merged at call time.

timeout (float, default: 30.0) : Maximum seconds to wait for the Actor run to finish.

max_retries (int, default: 3) : Number of retry attempts on failure. Set to 0 to prevent parallel duplicate runs on timeout.

Apify Store

Browse the Apify Store to find Actors for social media scraping, e-commerce data extraction, news aggregation, and more. Pass any Actor ID to actors to use it immediately.