Skip to content

The Agenterprise.ai DSL

DSL Levels Overview

Version 0.1.11

The Agenterprise DSL is structured in distinct layers as levels, each representing a core aspect of agentic, model-driven enterprise systems. The following example illustrates the main levels:

1. Environment Level

Defines the overall AI environment or context for the system.

ai_environment "AgentMicroservice"

2.1 Architecture Level

Describes the architecture, including environment IDs and technology stacks for services and AI components.

architecture{
    envid = "fb98001a0ce94c44ad091de3d2e78164"
    service-techlayer = aiurn:techlayer:github:www.github.com:agenterprise:service-layer-fastapi-base
    ai-techlayer = aiurn:techlayer:github:www.github.com:agenterprise:ai-layer-pydanticai
    data-techlayer = aiurn:techlayer:github:www.github.com:agenterprise:data-layer-pydantic     
}
Read on at Architecture Layer

2.2 Infrastructure Level

Specifies infrastructure resources such as LLMs, providers, endpoints, and versions.

infrastructure {
    llm "My LLM" {
        uid = aiurn:model:id:geepeetee
        provider = aiurn:model:provider:azure 
        model = "gpt-4o"
        endpoint = "https://any.openai.azure.com/"
        version = "2025-01-01-preview"
        aiurn:global:var:temperature = 0.7
        aiurn:global:var:costid = "ewe3949" 
        aiurn:global:var:hello = True 
    }
}
Read on at Infrastructure Layer

2.3 Datalayer

Defines Data-Entities used for functional layer

data{
    entity "Restaurant Query" {
        uid = aiurn:entity:id:restaurantquery
        element = aiurn:entity:var:question -> TEXT # "The question to the metre"
    }
    entity "Restaurant Answer" {
        uid = aiurn:entity:id:restaurantanswer 
        element = aiurn:entity:var:answer -> TEXT # "The answer of the metre"
        element = aiurn:entity:var:restaurant -> aiurn:entity:id:restaurant  # "The restaurant of the metre"
    }
    entity "Restaurant" {
        uid = aiurn:entity:id:restaurant 
        element = aiurn:entity:var:name -> TEXT # "The name of the restaurant"
        element = aiurn:entity:var:street -> TEXT # "The street where the restaurant is located"
        element = aiurn:entity:var:city -> TEXT # "The city where the restaurant is located"
        element = aiurn:entity:var:rating -> NUMBER # "The rating of the restaurant"
    }

    entity "BMI Query" {
        uid = aiurn:entity:id:bmiquery
        element = aiurn:entity:var:weight -> NUMBER # "The current weight of the person"
        element = aiurn:entity:var:height -> NUMBER # "The current height of the person in meters"
    }

    entity "BMI Result" {
        uid = aiurn:entity:id:bmiresult
        element = aiurn:entity:var:bmi -> NUMBER # "The calcualted bmi of the person"
    }
}
Read on at Data Layer

2.4 AI Functional Level

Defines agents, tools, and their properties, including prompts, references, variables, and endpoints.

functional{
        agent "Cook" {
            uid = aiurn:agent:id:cook
            namespace = aiurn:ns:moewe:kitchen
            systemprompt = "You're a four star rated metre working at restaurant https://moewe.agenterprise.ai/"
            llmref = aiurn:model:id:geepeetee 
            toolref = aiurn:tool:id:crawler:v2
            in = aiurn:entity:id:restaurantquery
            out = aiurn:entity:id:restaurantanswer 
            aiurn:global:var:name = "Max Mustermann"
            aiurn:global:var:role = "cook"
            aiurn:global:var:lifeycle = "permanent"
            aiurn:global:var:events = "onRestaurantOpening"

        }

        agent "Waiter" {
            uid = aiurn:agent:id:waiter
            namespace = aiurn:ns:moewe:guestroom
            systemprompt = "Du bist eine freundliche und aufmerksame Serviekraft und arbeitest im  Restaurant https://moewe.agenterprise.ai/"
            llmref = aiurn:model:id:geepeetee 
            toolref = aiurn:tool:id:bmi:v1
            toolref = aiurn:tool:id:crawler:v2
            aiurn:global:var:name = "Max Mustermann"
            aiurn:global:var:role = "waiter"
            aiurn:global:var:lifeycle = "permanent"
            aiurn:global:var:events = "onRestaurantOpening"
        }
        tool "bmicalculator" {
            uid = aiurn:tool:id:bmi:v1
            in = aiurn:entity:id:bmiquery
            out = aiurn:entity:id:bmiresult
            endpoint = "lambda bmiquery: ToolOutputType(bmi=round(bmiquery.weight / (bmiquery.height ** 2), 2))"
            type = aiurn:tooltype:code
            description = "Tool calculating the bmi by weight and height"

        }

         tool "Webcrawler" {
            uid = aiurn:tool:id:crawler:v2
            endpoint = "https://remote.mcpservers.org/fetch/mcp"
            type = aiurn:tooltype:mcp
            description = "Tool for fetching webpages"

        }

}

Read on at AI Functional Layer

Each level in the DSL enables clear separation of concerns, making it possible to model, generate, and manage complex agentic enterprise systems in a technology-neutral and flexible way.