Skip to content
SuperOptiX Logo

SUPEROPTIX AI

Full Stack Agentic AI Optimization Framework

Evaluation-First ⚡ Optimization-Core 🕸️ Multi-Agent Orchestration

Powered by DSPy. Refined by Superagentic AI.

Build once with SuperSpec and compile to your preferred framework.

Now supported: 🧪 RLM (Experimental) · 🗂️ StackOne Connectors · 🧬 GEPA Optimization

RLM Guide · StackOne Guide · Framework Support

What is SuperOptiX?

SuperOptiX is a universal agent optimization framework that lets you build, evaluate, and optimize agents across major frameworks with one workflow.

It keeps generated pipelines framework-native and readable in minimal mode, and enables full optimization lifecycle only when you opt in with --optimize.

Core Workflow

# Pull agent
super agent pull developer

# Compile minimal pipeline
super agent compile developer --framework dspy

# Run
super agent run developer --framework dspy --goal "Design a migration plan"

# Optional optimization path
super agent compile developer --framework dspy --optimize
super agent optimize developer --framework dspy --auto light

Why Teams Use It

🔬 Multi-Framework Freedom

  • Compile one SuperSpec into DSPy, OpenAI SDK, Claude SDK, Pydantic AI, CrewAI, Google ADK, DeepAgents
  • Microsoft support remains available in legacy mode
  • Switch frameworks without rewriting agent intent

🧬 Optimization by Default Path

  • Minimal runtime pipeline first
  • GEPA optimization loop when needed
  • BDD-style evaluation and repeatable quality checks

🗂️ Connector-Driven Agents

  • StackOne integration for SaaS tool access
  • Cross-framework connector compilation
  • Clear connector demos including Calendly flows

🧪 RLM Support

  • Experimental RLM support in active integrations
  • Fallback-friendly pipeline strategy
  • Unified sandbox support coming soon

Local and Cloud Routing

# Local Ollama
super agent run developer --framework dspy --local --provider ollama --model llama3.1:8b --goal "..."

# Cloud Google
super agent run developer --framework dspy --cloud --provider google-genai --model gemini-2.5-flash --goal "..."

Next Steps