← Back to Agents
Agent

Scientific Illustrator Agent

Transforms scientific text into structured, publication-grade visual prompts, diagrams, and figure concepts.

Designers: Nidhi Jha, Rahul Ramachandran

Accepts a research narrative and figure intent and produces figure specifications, captions, and image-generation prompts suitable for publication-grade scientific illustrations.

The Scientific Illustrator Agent is an AKD-designed cognitive assistant that transforms scientific text into structured, publication-grade visual representations, enabling scientists to communicate complex ideas with clarity and precision. A key feature of the agent is a faithfulness audit stage, which validates the transformation from input text to visual representation by explicitly distinguishing grounded facts, assumptions, and inferred elements before generating the final image prompt.

Inputs

Collaborative Design

The agent provides suggestions, while users retain control to finalize scientific accuracy, visual abstraction level and representation choices. Users define visualization intent with guided options:

Outputs

Pipeline

Text → Visual Logic → Faithfulness Audit → Traceable reasoning → Final Image Prompt → Image rendering (SVG/Image)