Research & Papers

SoftBoard uses multi-agent LLMs to automate low-fidelity prototyping

AI agents handle requirements, generate prototypes, and evaluate usability—all in one tool.

Deep Dive

SoftBoard, a web-based tool for low-fidelity prototyping in MVP development, integrates a prototype editor, team-based project organization, and a multi-agent system based on large language models. The system supports requirements elicitation and refinement, automates prototype generation, and evaluates interface quality using usability heuristics. This integration aims to reduce reliance on prior UX expertise and standardize the prototyping process. Aimed at software startups, the tool addresses gaps found in a systematic literature mapping: predominance of general-purpose solutions, lack of integrated methodological guidance, and limited use of AI. A feasibility study with software professionals is currently underway.

Key Points
  • Multi-agent LLM system handles requirements elicitation, prototype generation, and heuristic evaluation in one platform.
  • Targets software startups to reduce the need for dedicated UX expertise during MVP creation.
  • Accepted for publication at CBSoft 2026 (SBES-Tools), with a feasibility study underway.

Why It Matters

SoftBoard could democratize UX design for resource-constrained startups by automating expert-level prototyping and evaluation.

📬 Get the top 10 AI stories daily