Hello,
With the rapid growth of AI and GenAI workloads, it would be valuable for Diagrams to provide better support for AI-related services and architectural concepts.
Many AI platforms and services now have official branding and iconography available, and Diagrams could leverage custom icons for these services in the same way it supports many other cloud and platform offerings today.
In addition to service-specific icons, support for common AI concepts would help users create clearer architecture diagrams, including:
- Foundation models and LLMs (OpenAI GPT models, Anthropic Claude, Google Gemini, Meta Llama, Mistral, Cohere Command, Amazon Nova, DeepSeek, Qwen)
- AI agents and agent workflows (LangChain, LangGraph, CrewAI, AutoGen, Semantic Kernel)
- Retrieval and knowledge systems (OpenSearch, Haystack, LlamaIndex)
- Vector search and embeddings (Pinecone, Weaviate, Milvus, Qdrant, Chroma, pgvector, OpenAI Embeddings, Cohere Embed, Voyage AI)
- Model training and fine-tuning (Databricks Mosaic AI, Hugging Face, MLflow, Kubeflow)
- Model serving and inference (NVIDIA NIM, vLLM, Ollama, TensorRT-LLM, Hugging Face Inference Endpoints)
- Evaluation and safety components (LangSmith, Arize Phoenix, TruLens, NVIDIA NeMo Guardrails, Guardrails AI, Azure AI Evaluation)
While it is already possible to represent many AI services using custom icons, adding them directly to Diagrams would offer a better user experience. Built-in support would provide standardized icons, consistent naming, easier discoverability, and reduce the need for users to manage external assets when creating AI architecture diagrams.
I would like to work on this enhancement and contribute the required icons and node definitions.
Thanks!
Hello,
With the rapid growth of AI and GenAI workloads, it would be valuable for Diagrams to provide better support for AI-related services and architectural concepts.
Many AI platforms and services now have official branding and iconography available, and Diagrams could leverage custom icons for these services in the same way it supports many other cloud and platform offerings today.
In addition to service-specific icons, support for common AI concepts would help users create clearer architecture diagrams, including:
While it is already possible to represent many AI services using custom icons, adding them directly to Diagrams would offer a better user experience. Built-in support would provide standardized icons, consistent naming, easier discoverability, and reduce the need for users to manage external assets when creating AI architecture diagrams.
I would like to work on this enhancement and contribute the required icons and node definitions.
Thanks!