Rapid

Overview

An AI-powered prototype generator that transforms client problems described in plain language into complete prototype briefs in under two minutes. Generates problem framing, solution concepts, interactive wireframes, tech stack recommendations, and hero images — then supports iterative refinement through conversation.

Key Features

• Two-stage AI generation pipeline: Claude Sonnet for solution concepts and Claude Haiku for wireframe rendering with temperature-tuned prompts
• Interactive wireframe rendering from 4 template archetypes (analytics dashboard, operations console, app portal, document workflow) via sandboxed iframe
• Hero image generation using OpenAI with smart product-type detection and industry-specific palette derivation
• Iterative refinement through conversational input — AI updates affected sections without regenerating the full brief
• 8 pre-baked fixture demos with realistic phased generation for instant showcase without API calls

Challenges & Solutions

The project tackled challenges including building a two-stage AI generation pipeline with Claude tool_use for structured JSON output, rendering self-contained wireframe HTML in sandboxed iframes without external resources, implementing smart product-type detection via regex for contextual hero image generation with graceful fallback handling, managing session state through a useReducer state machine supporting idle through complete phases with refinement iterations, and achieving 179 passing tests across utilities, reducers, prompts, templates, fixtures, and API routes.