Progressive AI Architecture

🏗️ Agent Tiers

Scale from simple oracles to autonomous sovereigns. Each tier builds upon the previous, ensuring a smooth development journey with comprehensive features and capabilities.

5 Progressive Tiers

DSPy Integration

Production Ready

🚀 Tier Overview & Pricing

Progressive complexity from simple oracles to autonomous sovereigns. Each tier builds upon the previous, ensuring a smooth development journey. Start free with free tiers and scale to enterprise solutions.

Tier	Status	Pricing	Complexity	Use Case
🧙‍♂️ Oracles Entry Level - Free Trial	Available	Free forever	Low	Basic automation, simple Q&A
🧞‍♂️ Genies Intermediate	Available	Free to explore. Pro tools optional	Medium	Customer service, content creation
🎭 Protocols Advanced	Available	Custom pricing based on use case	High	Business processes, decision making
🤖 Superagents Expert	Coming Soon	Tailored enterprise deployment	Expert	Complex business workflows
👑 Sovereigns Autonomous	Future	Invite-only / Strategic collaboration	Autonomous	Large-scale AI operations

Progressive complexity and capabilities

Each tier builds upon the previous

Scale as your needs grow

🧙‍♂️ Oracles

Entry Level - Free Trial

Simple, fast question answering systems with LLM interaction

Available

Free forever

Single-step reasoning

Template-based responses

Built-in optimization and validations

Any LLM Support

Model Management

Few Shot Optimization

Simple Evals

BDD Spec Runner

Simple Sequential Multi Agent Orchestra

Static Pipelines Code with SuperOptiX DSPy Mixin

Demo Purpose Outputs

Basic tracing and observability

Examples:

FAQ bots

Data formatters

Simple Q&A systems

Text processors

Basic chatbots

Information retrieval

oracles_agent.yaml

# oracle_agent.yaml
apiVersion: agent/v1
kind: AgentSpec
metadata:
  name: faq-bot
  tier: oracle
 
spec:
  language_model:
    provider: openai
    model: gpt-4o
 
  persona:
    role: "FAQ Assistant"
    description: "Answers common questions"
 
  tasks:
    - name: "answer_question"
      description: "Provide helpful answers"
 
  optimization:
    type: "basic"
    metrics: ["accuracy", "response_time"]
 
  evaluation:
    type: "simple"
    test_cases: 10

🧞‍♂️ Genies

Intermediate

Multi-step reasoning agents with tools and memory

Available

Free to explore. Pro tools optional

Multi-step reasoning with ReAct

Dynamic tool selection and usage

Memory integration and learning through RAG

Function calling LLM Support

Custom Function calling DSPy tools

RAG with favorite vectorDB Support

Model Management with MLX, HF, Ollama and LM Studio

Few Shot and Labeled Few Shot Optimization

Simple Evals

Basic DSPy Memory Support

BDD Spec Runner basic metrics

Sequential Multi Agent Orchestra

Static Pipelines Code with SuperOptiX DSPy Mixin

Demo Purpose Outputs with usage tracking

Basic Tool Tracing Observability and Tool call

Multi-Agent Orchestra with demo outputs

Examples:

Customer support agents

Content writers

Research assistants

Data analysts

Code assistants

Documentation generators

genies_agent.yaml

# genie_agent.yaml
apiVersion: agent/v1
kind: AgentSpec
metadata:
  name: support-agent
  tier: genie
 
spec:
  language_model:
    provider: openai
    model: gpt-4o
 
  tools:
    - name: "knowledge_base"
      type: "rag"
      source: "company_docs"
      vector_db: "chromadb"
    - name: "ticket_system"
      type: "api"
      endpoint: "https://api.company.com/tickets"
 
  memory:
    type: "conversation"
    max_tokens: 4000
    vector_store: "redis"
 
  reasoning:
    type: "react"
    max_steps: 5
 
  optimization:
    type: "advanced"
    metrics: ["accuracy", "response_time", "user_satisfaction"]

🎭 Protocols

Advanced

Advanced agents with industry protocols and complex workflows

Available

Custom pricing based on use case

Advanced agents with MCP & A2A protocols

Integrate with external APIs, systems, workflows

Everything from Oracles and Genies

Custom Function calling DSPy tools

Agentic RAG with popular vectorDB Support

AgentVectorDB Integration

Advanced Model Management with vLLM, SGLang, TGI servers for Production deployment

Advanced DSPy and Custom Optimizers

Layered Memory Support

Automated Basic Synthetic Data Generation

BDD Spec Runner with advanced metrics and validations

Parallel Multi Agent Orchestra

Controlled DSPy Pipelines (No Mixin)

Production Worthy Agent Output format suitable for multi-agent system

Advanced Tracing Observability and Tool

Integration with third party tools like MLflow

Basic Planner → Executor Multi Agent Orchestra

Basic Kubernetes Style Orchestra

Examples:

Sales qualification

Risk assessment

Compliance monitoring

Workflow automation

Enterprise integrations

Complex business processes

protocols_agent.yaml

# protocol_agent.yaml
apiVersion: agent/v1
kind: AgentSpec
metadata:
  name: sales-qualifier
  tier: protocol
 
spec:
  protocols:
    - name: "MCP"
      version: "1.0"
    - name: "A2A"
      version: "1.0"
 
  coordination:
    type: "hierarchical"
    leader: "sales-manager"
 
  integrations:
    - name: "crm"
      type: "salesforce"
    - name: "email"
      type: "outlook"
    - name: "calendar"
      type: "google_calendar"
 
  memory:
    type: "layered"
    layers: ["conversation", "knowledge", "long_term"]
 
  optimization:
    type: "custom"
    optimizer: "enterprise_optimizer"
    metrics: ["conversion_rate", "response_time", "accuracy"]

🤖 Superagents

Expert

Multi-agent systems with coordination where a lead agent called Superagent may spawn automated subagents and work with other superagents.

Coming Soon

Tailored enterprise deployment

Superagents orchestrating other agents

AgentLines for scalable multi-agent governance

Everything from Oracles, Genies and Protocols

Agentic DSPy Pipeline for Superagent

Advanced Model Management with vLLM, SGLang, TGI servers for Production deployment

Integration with high level GPU infra and MLOps tools for deployment

Combination of LLM and Fine Tuned SLMs

Context Management with VectorDBs and Advanced Memory

Agentic BDD Spec Runner within orchestra and AgentLines

Human in the loop interaction based on defined criteria

Integration with third party DevOps, MLOps Cloud providers

Examples:

E-commerce platforms

Research teams

Trading systems

Content studios

Enterprise platforms

AI-powered companies

superagents_agent.yaml

# superagent_orchestra.yaml
apiVersion: agent/v1
kind: OrchestraSpec
metadata:
  name: ecommerce-platform
  tier: superagent

spec:
  superagents:
    - name: "product-manager"
      role: "coordinator"
      agents: ["researcher", "writer", "reviewer"]
    - name: "customer-service"
      role: "specialist"
      agents: ["faq-bot", "escalation-handler"]
      
  infrastructure:
    gpu: true
    scaling: "auto"
    monitoring: "comprehensive"
    
  coordination:
    type: "hierarchical"
    communication: "agent_lines"
    
  optimization:
    type: "distributed"
    strategy: "multi_agent_optimization"

Enterprise Feature

This tier is available for enterprise customers. Contact us to learn more about implementation details.

👑 Sovereigns

Autonomous

Autonomous AI systems suitable for large-scale AI operations and enterprise workflows with the highest level of AI autonomy.

Future

Invite-only / Strategic collaboration

Advanced multi-agent orchestration

Strategic planning and execution

Automatic discovery of agents based on task or goal

Ephemeral Agents making decisions and handling tasks

Integration with agent marketplace for choosing agents for tasks

Multiple LLM and Fine Tuned SLMs

Context Management with VectorDBs and Advanced Memory

Agentic BDD Spec Runner within orchestra and AgentLines

Integration with Multiple third party DevOps, MLOps Cloud providers

Examples:

AI-powered companies

Research labs

Autonomous systems

Strategic advisors

Self-managing platforms

Autonomous organizations

sovereigns_agent.yaml

# sovereign_system.yaml
apiVersion: agent/v1
kind: SovereignSpec
metadata:
  name: ai-company
  tier: sovereign

spec:
  autonomy:
    level: "full"
    decision_making: "strategic"
    self_improvement: true
    
  discovery:
    agent_discovery: true
    task_decomposition: true
    resource_allocation: true
    
  governance:
    type: "constitutional"
    safety_measures: "comprehensive"
    human_oversight: "minimal"
    
  optimization:
    type: "autonomous"
    self_improvement: true
    strategic_planning: true

Enterprise Feature

This tier is available for enterprise customers. Contact us to learn more about implementation details.

Ready to Choose Your Tier?

Start with Oracles for simple tasks and scale up as your needs grow. Each tier builds upon the previous, ensuring a smooth development journey with comprehensive features and capabilities.