Under the hood
How BA Assistant Turns Raw Requirements Into a Complete BA Report
A 5-agent workflow analyzes your input from multiple angles: BA analysis, product backlog, architecture, risks, QA, and diagrams.
1. What BA Assistant Does
You paste raw requirements. Five specialized agents analyze the input and produce a structured, delivery-ready BA/PO report with all the artifacts a Product Owner or Business Analyst needs.
2. The 5-Agent Workflow
1. BA Requirements Analyst
Structures raw needs into clear functional & non-functional requirements. Extracts entities, business rules, and constraints.
deepseek-v4-flash
2. Product Owner Analyst
Converts requirements into epics, features, and user stories with MoSCoW prioritization and backlog structure.
deepseek-v4-flash
3. Solution Architect
Designs technology stack, integration patterns, data models, and NFRs with deployment strategy recommendations.
deepseek-v4-flash
4. Risk & QA Reviewer
Identifies technical, business, and project risks. Validates completeness and cross-agent output consistency.
deepseek-v4-flash
5. Mermaid Diagram Designer
Generates process flows, sequence diagrams, state charts, and data flow diagrams — paste-ready for any markdown document.
deepseek-v4-flash
3. Request Flow
1. User pastes raw requirements →
2. BA Analyst structures & categorizes →
3. PO Analyst extracts features & stories →
4. Architect designs system & integrations →
5. Risk & QA reviews all outputs →
6. Diagram Designer generates Mermaid charts →
7. Consolidated report delivered to user
4. What the Report Includes
- Functional requirements, categorised and traceable
- Non-functional requirements (performance, security, scalability)
- Features with MoSCoW priority, dependencies, and complexity
- User stories with Given-When-Then acceptance criteria
- Architecture: stack, patterns, integrations, data model
- Risk matrix with probability, impact, and mitigations
- Mermaid diagrams (flowchart, sequence, state)
- Executive summary and implementation roadmap
⚠️ Important: BA Assistant is designed for early-stage analysis and documentation support. AI-generated outputs should always be reviewed by a human before use in production delivery. Do not paste confidential or personally identifiable information.
5. Running Costs
| Service | Role | Plan | Cost |
| DeepSeek API | 5-agent AI engine | Pay-per-use | ~$0.01 / analysis |
| Streamlit Cloud | App hosting | Free tier | $0 |
| GitHub Pages | Landing page + docs | Free | $0 |
| Cloudflare | Domain + DNS | Registrar | ~$10 / year |
6. Tech Stack
| Layer | Technology |
| Agent Framework | Agno (Python) |
| LLM Provider | DeepSeek V4 Flash |
| App UI | Streamlit |
| App Hosting | Streamlit Cloud |
| Landing Page | Static HTML/CSS on GitHub Pages |
| Domain & DNS | Cloudflare |