Open Source Tools for Modern AI Development

Build powerful AI applications with our suite of open-source packages and enterprise-grade services. From LLM integration to agent coordination, Dhenara provides everything you need.

Get started Documentation

python

from dhenara.agent.dsl import PLACEHOLDER, AIModelNode, AIModelNodeSettings, EventType, FlowDefinition
from dhenara.ai.types import AIModelCallConfig, Prompt

test_mode = False
models_with_options = {
    "gpt-image-1": { "quality": "medium", "size": "1536x1024", "n": 2},
    "imagen-3.0-generate": { "aspect_ratio": "16:9", "number_of_images": 1, "person_generation": "dont_allow"},
    "imagen-3.0-fast-generate": { "aspect_ratio": "16:9", "number_of_images": 1, "person_generation": "dont_allow"},
}

single_img_flow = FlowDefinition()
single_img_flow.vars({"task_spec": PLACEHOLDER})
single_img_flow.node(
    "image_generator",
    AIModelNode(
        models=list(models_with_options.keys()),
        pre_events=[EventType.node_input_required],
        settings=AIModelNodeSettings(
            system_instructions=[ "You are a professional image generation agent."],
            prompt=Prompt.with_dad_text("$expr{task_spec.prompt}. Create image"),
            model_call_config=AIModelCallConfig( options={}),
            save_generated_bytes=True,
            bytes_save_path=global_image_directory,
            bytes_save_filename_prefix="$expr{task_spec.filename}",
            sleep_after=20,
        ),
    ),
)

multi_image_flow = FlowDefinition()
multi_image_flow.vars({"image_task": image_task})
multi_image_flow.for_each(
    id="image_gen_loop",
    statement="$expr{image_task.task_specifications}",
    item_var="task_spec",
    max_iterations=20,
    body=single_img_flow,
)

Built for developers

Everything you need to build AI-powered applications

Our tools are designed to make AI development faster, more flexible, and more reliable, whether you're building a simple application or a complex enterprise system.

Open Source Core: Our core libraries are fully open-source, allowing for community contributions and transparency.
Provider Agnostic: Work with models from OpenAI, Google AI, Anthropic, and other providers through a consistent interface.
Performance Focused: Built for high-performance applications with efficient model integration and streaming support.
Enterprise Ready: Secure, scalable, and ready for enterprise deployment with comprehensive monitoring and support.
Dynamic Model Selection: Switch between models and providers at runtime without reconfiguration or code changes.
Type Safety: Built with Pydantic for robust type checking and validation throughout your application.