OpenCommerce Docs
A complete reference for agentic AI commerce and intelligent commerce — from language model fundamentals to standard protocols, knowledge and memory systems, development frameworks, and multi-agent architectures. Every page includes a full explanation, code samples, and an architecture diagram.
Foundations
Large Language Model (LLM) Fundamentals
How large language models work: tokenization, transformer architecture, training, and limitations.
FoundationsAI Agents & Agentic AI
What sets an agent apart from a plain language model, its core components, and the Reason-Act loop.
Protocols
Model Context Protocol (MCP)
Securely connecting AI agents to a business's internal tools and data.
UCPUniversal Commerce Protocol (UCP)
An open standard for discovery, search, and transactions by shopping agents.
ProtocolsAgent-to-Agent Protocol (A2A)
An open protocol for direct collaboration between independent agents from different vendors.
GEOGenerative Engine Optimization (GEO)
Optimizing content to be surfaced in the answers of generative AI engines.
Knowledge & Memory
Retrieval-Augmented Generation (RAG)
Injecting external knowledge into a language model's answers via semantic document retrieval.
Knowledge & MemoryAgentic RAG
When the agent itself decides when, how, and how many times to retrieve.
Knowledge & MemoryAgent Memory Systems
Short-term, long-term, episodic, and procedural memory in AI agents.
Frameworks
LangChain
An open-source framework for building LLM chains: prompts, models, tools, and retrievers.
FrameworksLangGraph
Agent orchestration as a state graph, with support for loops and conditional branches.
Multi-Agent
Multi-Agent Systems
Why and when several specialized agents outperform one do-everything agent.
Multi-AgentMulti-Agent Architecture Patterns
Supervisor, Hierarchical, Peer-to-Peer, and Pipeline patterns.
Multi-AgentAgent Orchestration
Task decomposition, routing, result aggregation, and error handling at the orchestrator level.
Iran Market
Why Agentic Commerce Is a Revolutionary Opportunity for Iran
Why Iran's e-commerce infrastructure is well-suited to leapfrog into agentic commerce.
Iran MarketFitting UCP/MCP to Iran's Payment Infrastructure
How UCP and MCP align with Shetab, Shaparak, and domestic payment gateways.
Iran MarketRegulatory Recommendations for Iranian Lawmakers
A proposed framework for liability, transparency, privacy, and a regulatory sandbox.
Iran MarketAn Agentic Commerce Adoption Roadmap for Iranian Businesses
A practical five-stage roadmap, from a simple file to full integration with shopping agents.
Where should I start?
If you're new to agentic AI commerce concepts, start with Foundations. If you're looking to implement this directly on your own site, we recommend following this order: first, implement GEO on your existing content so agents can discover you at all, then set up UCP to enable transactions, and finally, for controlled access to sensitive internal data, look at MCP. For an instant start with no manual coding, use the free UCP generator.
The OpenCommerce blog also has narrative, example-driven guides, and the Sandbox is an environment for running live UCP, MCP, and GEO code samples directly in your browser.