What is Upsonic?
Upsonic is an open-source AI agent framework designed for building production-ready AI agents and digital workers. It provides a comprehensive toolkit for creating intelligent agents that can interact with external tools, manage memory, enforce safety policies, and coordinate with other agents.Upsonic supports multiple AI providers including OpenAI, Anthropic, Azure, AWS Bedrock, Google, and more through a unified interface.
Key Features
Multi-Provider Support
Use OpenAI, Anthropic, Azure, Bedrock, Google, and other providers through a unified API
Safety Engine
Policy-based content filtering for user inputs, agent outputs, and tool interactions
Autonomous Agent
Pre-configured agent with filesystem and shell capabilities for coding tasks
Memory Management
Session memory, chat history, and user profile extraction with multiple storage backends
Multi-Agent Teams
Coordinate multiple agents in sequential or parallel workflows
Tool Integration
MCP tools, custom tools, and human-in-the-loop workflows
OCR Support
Extract text from images and PDFs using multiple OCR engines
Production Ready
Monitoring, metrics, enterprise deployment, and cost tracking
Use Cases
Document Analysis
Extract and process text from images and PDFs with built-in OCR support for invoice processing, form extraction, and document understanding.Customer Service Automation
Build agents with memory and session context to provide personalized customer support across multiple conversations.Financial Analysis
Create specialized agents that analyze market data, generate reports, and provide insights using tools like YFinance.Compliance Monitoring
Enforce safety policies across all agent interactions to ensure content meets regulatory and business requirements.Research & Data Gathering
Automate research workflows with multi-agent collaboration, web search tools, and knowledge base integration.DevOps Automation
Use AutonomousAgent to read, write, and execute code in sandboxed workspaces for automated DevOps tasks.Architecture Overview
Upsonic follows a modular architecture with clear separation of concerns:Core Components
Agent
The main orchestrator that manages task execution, tool calling, and model interactions. Supports both synchronous (
do) and asynchronous (do_async) execution.Task
Defines what the agent should accomplish, including description, tools, response format, and attachments (files/images).
Model Provider
Abstraction layer for different AI providers. Supports reasoning modes, thinking tools, and provider-specific features.
Tools
Extend agent capabilities with external tools. Supports functions, tool kits, MCP servers, and agent-as-tool patterns.
Memory
Persist conversation history and user profiles across sessions using various storage backends (in-memory, SQLite, Redis, PostgreSQL, MongoDB).
Why Upsonic?
Production-Ready Out of the Box
Production-Ready Out of the Box
Unlike lightweight frameworks, Upsonic includes monitoring, cost tracking, safety policies, and error handling - everything you need for production deployments.
Provider Flexibility
Provider Flexibility
Switch between OpenAI, Anthropic, Azure, or Bedrock with a single line of code. No vendor lock-in.
Safety First
Safety First
Built-in safety engine with pre-configured policies for PII protection, content filtering, and compliance enforcement.
Autonomous Capabilities
Autonomous Capabilities
AutonomousAgent provides filesystem and shell access in sandboxed environments - perfect for coding assistants and DevOps automation.
Rich Ecosystem
Rich Ecosystem
Integrate with hundreds of MCP tools, use pre-built tool kits (YFinance, Tavily, DuckDuckGo), or create custom tools.
Quick Example
Here’s a simple example to get a feel for Upsonic:Community and Support
- Documentation: docs.upsonic.ai
- GitHub: github.com/Upsonic/Upsonic
- Issue Tracker: Report bugs and request features
- Changelog: See what’s new in each release