idea-generation

AI-powered idea generation uses configurable frameworks like pain point or market gap analysis to produce scored ideas. This skill generates ideas for specified domains, evaluates them against weighted criteria such as market size and feasibility, and applies semantic duplicate detection to surface only the most viable concepts.

Yes, the generation pipeline supports multiple domains and frameworks. You configure the domain (Fintech, Education, etc.), select a generation framework, and the system produces AI-generated ideas with real-time progress logs via SSE, then ranks them by your chosen evaluation criteria.

The skill uses embedding-based similarity checks to prevent redundant ideas. After AI generation, ideas are compared semantically using embeddings stored in the IdeaRepository, ensuring only novel concepts pass through deduplication filters.

The skill supports configurable generation frameworks including pain point identification, market gap analysis, and unbundling strategies. You select the framework in the YAML configuration, and the PromptBuilder structures the AI call to generate ideas aligned with that approach.

Real-time logging is powered by Server-Sent Events (SSE). As the generation pipeline processes ideas through callAI(), parseAIResponse(), and scoring stages, live logs stream progress, giving you transparency and control over the entire workflow.

Ideas are dynamically scored against weighted criteria you define—such as market size, competition level, and technical feasibility. The scoring system applies these weights to produce objective rankings, allowing you to identify the most promising business opportunities quickly.