soe-ai 0.1.1__py3-none-any.whl → 0.1.2__py3-none-any.whl

soe/docs/advanced_patterns/index.md

@@ -0,0 +1,49 @@
+# SOE Advanced Patterns
+
+Beyond basic workflows, SOE enables powerful patterns for building intelligent systems.
+
+---
+
+## The Patterns
+
+| Pattern | Description | Use Case |
+|---------|-------------|----------|
+| [Configuration & Context Inheritance](guide_inheritance.md) | Inherit config and context across executions | Workflow chaining, multi-phase pipelines |
+| [Fan-Out & Aggregations](guide_fanout_and_aggregations.md) | Parallel processing and data collection | Distributed work, map-reduce |
+| [Self-Evolving Workflows](self_evolving_workflows.md) | Workflows that modify themselves at runtime | Adaptive systems, hot-patching |
+| [Swarm Intelligence](swarm_intelligence.md) | Multiple agents voting and reaching consensus | Jury systems, quality control |
+| [Hybrid Intelligence](hybrid_intelligence.md) | Mixing deterministic logic with AI | Safety rails, enterprise compliance |
+
+---
+
+## Philosophy
+
+These patterns share a common principle: **SOE is a protocol, not a product**.
+
+- **Self-Evolving**: YAML is text. LLMs can generate text. Therefore, LLMs can generate workflows.
+- **Swarm**: Signals are cheap. Counting signals is cheaper than parsing paragraphs.
+- **Hybrid**: Routers are deterministic. LLMs are creative. Combine them.
+
+---
+
+## Future Patterns
+
+The following patterns are possible with SOE but not yet documented:
+
+### Fractal Orchestration
+Child workflows spawning child workflows, ad infinitum. A "CEO" workflow spawns "Manager" workflows, which spawn "Worker" workflows.
+
+### Universal Backend
+Any workflow can be wrapped as a tool for another workflow. A "Research Agent" workflow becomes a tool for a "Writer Agent" workflow.
+
+### Edge AI
+SOE has no heavy dependencies. It can run on Raspberry Pi with local LLMs (Llama.cpp), or in the browser via WebAssembly.
+
+---
+
+## Getting Started
+
+1. Study the basic [User Guide](../index.md) first
+2. Read [Self-Evolving Workflows](self_evolving_workflows.md) to understand runtime modification
+3. Try [Swarm Intelligence](swarm_intelligence.md) for multi-agent coordination
+4. Apply [Hybrid Intelligence](hybrid_intelligence.md) for production safety