ArchForge AI

AI system architecture advisor by Sharad Khare.

ArchForge AI is a dual-model architecture copilot that combines DeepSeek R1 reasoning with Claude synthesis to produce structured technical analysis, implementation roadmaps, and Mermaid-ready architecture documentation.

Built and maintained by Sharad Khare — AI strategist, full-stack developer, and creator of practical agent workflows.

---

What this project does

ArchForge turns a natural-language project brief into a full architecture package:

1. Reason — DeepSeek R1 analyzes requirements and produces chain-of-thought reasoning

2. Structure — DeepSeek outputs a JSON technical analysis (patterns, infra, security, compliance)

3. Synthesize — Claude explains decisions, builds a roadmap, and writes a spec with Mermaid diagrams

The result is an actionable architecture report you can share with engineering teams, stakeholders, or investors.

---

Why Sharad Khare built ArchForge AI

System design decisions are often opaque — teams need both rigorous reasoning and readable documentation. ArchForge demonstrates a practical multi-model pattern: one model for deep structured reasoning, another for polished technical communication.

---

How it works

User Brief
   │
   ▼
DeepSeek R1 (deepseek-reasoner)
   ├── Chain-of-thought reasoning
   └── Structured JSON technical analysis
   │
   ▼
Claude 3.5 Sonnet (via Agno)
   ├── Decision explanations
   ├── Implementation roadmap
   └── Markdown spec + Mermaid diagrams

Analysis coverage

| Area | Output |

|------|--------|

| Architecture | Pattern choice, trade-offs, cost estimates |

| Infrastructure | Resources, scaling policies, specs |

| Security | Controls, compliance mapping, data classification |

| Database | SQL / NoSQL / hybrid recommendations |

| Performance | Metrics, targets, priorities |

| Risk | Assessment and mitigation strategies |

---

Features

• Dual-model reasoning pipeline (DeepSeek + Claude)
• Structured JSON schema for architecture decisions
• Expandable reasoning and analysis panels in Streamlit
• Chat history with session persistence
• Mermaid.js diagram descriptions in final report
• HIPAA, GDPR, SOC2, ISO27001 compliance awareness

---

Quick start

cd archforge-ai
pip install -r requirements.txt
streamlit run app.py

Required API keys

• DeepSeek API key — for R1 reasoning and structured analysis
• Anthropic API key — for Claude architecture synthesis

---

Example prompts

• "Design a HIPAA-compliant healthcare data platform for 10,000 users with $50k budget"
• "Architect a fintech payment system handling 1M transactions/day with SOC2 compliance"
• "Plan a serverless e-commerce backend for a startup with rapid growth expectations"
• "Design an event-driven analytics pipeline integrating HL7/FHIR hospital data"

---

Project structure

archforge-ai/
├── app.py
├── archforge/
│   ├── config.py
│   ├── schemas.py
│   ├── prompts.py
│   └── chain.py
├── requirements.txt
├── pyproject.toml
└── README.md

---

Who this is for

• Software architects and senior engineers
• Startup founders planning technical infrastructure
• Consultants producing system design documents
• Developers learning multi-model agent patterns

Use cases

• Turn project briefs into structured architecture analysis
• Compare microservices vs monolithic vs serverless trade-offs
• Generate compliance-aware security and infrastructure plans
• Produce stakeholder-ready specs with implementation roadmaps

---

Author

• Website: sharadkhare.in

License

MIT © Sharad Khare