Skip to content

XWhy explainers

This page is the authoritative capability map for the current documentation release.

Explainer area Input or system context Intended explanation task Status
Image Classification Image Local image-region influence on classification Available
Image Generation & Editing Prompt, optional source image, and other model conditioning Prompt and conditioning influence on generated or edited images Available
LLM Prompt and generated response Local prompt influence on response behaviour Available
Tabular Structured features Feature influence on classification and regression Available
Text Text input Word-level influence on conventional text-model predictions Available
Point Cloud 3D points or point groups Local influence on 3D prediction Available
Time Series Ordered observations Influence of observations, windows, and temporal patterns Coming soon
Multimodal Two or more modalities Modality-specific and cross-modal contributions Coming soon
Agentic AI Agent trajectory including plans, retrieval, memory, tools, states, and actions Explain decisions, actions, state transitions, and uncertainty or failure propagation within an autonomous agent Coming soon
Multi-Agent AI Interacting agents, messages, roles, dependencies, and shared state Explain agent contribution, communication, coordination, disagreement, and uncertainty or failure propagation across agents Coming soon

The current generative-image API is ImageGenerationAndEditingExplainer. The Image Generation and Editing Examples page describes comprehensive examples for generating and editing images using OpenAI, ByteDance, Gemini, and other compatible providers.

Agentic AI and Multi-Agent AI are currently research-roadmap areas rather than exported explainers. Their documentation defines intended explanation targets and evaluation principles without implying that executable implementations already exist.

Status definitions

Available means an executable implementation and dedicated guide exist in the repository.

Under construction means a public interface is present, but a supported end-to-end workflow is not yet available.

Coming soon means the documentation reserves a stable location for a planned capability that is not currently implemented.