Overview: Upsonic Framework

Title

Upsonic Framework - A Reliability-Focused Framework for Production-Ready AI Agents

How to Use

1. Installation: Install the Upsonic package via pip:

   pip install upsonic

2. Set API Key: Configure your OpenAI or Anthropic API key:

   export OPENAI_API_KEY=sk-***

3. Basic Usage: Create tasks and agents to execute them:

   from upsonic import Task, Agent
   task = Task("Who developed you?")
   agent = Agent("Coder")
   agent.print_do(task)

Key Features

• Reliability Layer: Ensures accurate outputs with verifier and editor agents, verification rounds, and loops.
• Model Context Protocol (MCP): Supports hundreds of pre-built tools for seamless integration.
• Production-Ready Scalability: Deployable on AWS, GCP, or locally using Docker.
• Task-Centric Design: Supports basic LLM tasks, advanced V1 agents, and complex V2 agents with MCP.
• Computer Use Integration: Execute human-like tasks using Anthropic’s ‘Computer Use’ capabilities.
• Tool-Calling Server: Robust server API interactions with exception-secure tool management.
• Structured Outputs: Define exact response formats using Pydantic BaseClass.

Use Cases

1. Web Content Analysis: Fetch and analyze web pages using MCP tools.

   web_agent = Agent("Web Content Analyzer", model="openai/gpt-4o")
   task = Task(description="Fetch and analyze content from a URL.", context=["https://upsonic.ai"])
   result = web_agent.print_do(task)

2. Multi-Agent Task Distribution: Collaborate across agents for complex problem-solving.

   researcher = Agent("Company Researcher", company_url="https://redis.io/")

3. Secure Runtime: Run agents in an isolated environment for enhanced security.
4. Business Applications: Generate reports, analyze competitors, and automate workflows with high reliability.