Whiteboard to Requirements

How It Works

Upload a photo of your whiteboard, hand-drawn process flow, or system diagram. The AI reads the sketch using vision models, asks 3-5 clarifying questions about actors, triggers, and system boundaries, then generates a complete set of structured BA artifacts — all in under 60 seconds.

1. Upload Sketch

Photo of your whiteboard, hand-drawn flow, or system diagram. The app accepts common image formats — just drag and drop.

2. AI Analyzes

Gemini 3.5 Flash reads the sketch using vision capabilities and generates 3-5 clarifying questions to better understand the process, actors, and intent behind the drawing.

3. You Clarify

Answer the questions about actors, triggers, and system boundaries. Your answers give the AI the context it needs to produce accurate, relevant artifacts.

4. Get Artifacts

In under 60 seconds, you get INVEST user stories, functional requirements with FR-XXX IDs, Mermaid sequence and flowchart diagrams, and a gap analysis — all rendered live in the browser.

What You Get

📋 Epic Summary

2-3 sentence product capability overview for stakeholders. Distills the whiteboard sketch into a clear, concise summary anyone can understand.

🔧 Functional Requirements

FR-XXX IDs grouped by module with testable requirement statements. Each requirement is traceable and ready for your specification document.

✅ INVEST User Stories

Proper format with acceptance criteria — ready for sprint planning. Stories follow the INVEST framework: Independent, Negotiable, Valuable, Estimable, Small, Testable.

🔄 Mermaid Diagrams

Sequence diagram and flowchart rendered live in the browser. Copy the Mermaid code directly into your documentation or export as SVG.

🔍 Gap Analysis

Ambiguities, missing context, and assumptions flagged for clarification. Helps you identify what's still unclear before you commit to implementation.

📥 Download as Markdown

Export everything as a .md file for your repo or documentation tool. Drop it into Confluence, Notion, or your PRD with zero reformatting.

Tech stack

Layer	Technology
UI Framework	🎨 Streamlit
Agent Framework	🧠 Agno — multi-agent orchestration
Vision Model	👁️ Gemini 3.5 Flash — reads whiteboard sketches, generates clarifying questions
Generation Model	⚡ DeepSeek V4 Flash — structured output generation (user stories, FRs, diagrams)
Diagrams	📊 Mermaid.js — sequence diagrams and flowcharts rendered client-side
Backend	🐍 Python — orchestration and artifact generation
Hosting	☁️ Hugging Face Spaces — free hosting with auto-deploy from GitHub

Open source on GitHub. Two-model architecture: Gemini handles vision and questions, DeepSeek generates structured outputs — keeping it fast and cost-efficient.