Samvada-Studio Diagrams

A backend engineer's journey from collecting features over months to shipping a production-ready LLM chat UI with AI assistance

By Dhruvin Soni | January 2026 | 15 min read

🔥 The Frustration That Started It All

When ChatGPT launched in late 2022, I tried every LLM chat UI I could find.

Each had brilliant features—but they were scattered:

• Gemini had beautiful inline editing
• ChatGPT had useful archiving
• Copilot had smooth keyboard shortcuts

I wanted them all. In one place.

Over the next few months, I collected features I wanted in my ideal LLM chat UI. I scribbled notes, saved screenshots, and built a comprehensive list of 20+ features.

Then, with modern AI-assisted development tools, I turned that vision into reality in just one week.

I had collected features over months. With AI assistance, implementation took 1 week.

Today(23-Jan-2026), I shipped Samvada Studio with:

• ✅ 37+ features (20 original + 17 bonus)
• ✅ 6 LLM providers (OpenAI, Anthropic, Google, Ollama, Azure, Custom)
• ✅ 15,000+ lines of documentation
• ✅ 100% local-first (no backend, no tracking)
• ✅ Security-first architecture (API keys in memory only)