LumiChats was built by two people: Aditya Kumar Jha (the technical co-founder who built the platform) and Shikhar Burman (the product and content co-founder who shaped the user experience). Both were students at the time. Both had the same problem.
The Problem They Were Solving
In 2025, Aditya was paying for ChatGPT Plus, a Claude subscription, and an API key for personal projects — a total of roughly ₹5,000–₹6,000 per month across platforms he used heavily during project seasons and barely touched during breaks. The money drained out every month regardless of whether he was in an intensive study phase or on a two-week vacation. The subscription model is designed for consistent daily users. Most students are not consistent daily users. They are burst users.
Shikhar had the same observation from the content side: every 'affordable AI for students' guide he found was written for Western students with Western salaries, recommended tools priced at $20/month, and assumed that ₹1,800 per month was a trivial spend. For most Indian students — and for most students globally — ₹1,800/month for a single AI subscription is a significant discretionary expense. The market needed something built for how students actually use AI.
What They Built
The technical architecture of LumiChats is deliberately ambitious for its price point. Every AI call is routed through OpenRouter with an 8-key load-balanced pool and automatic fallback chains, giving users access to 40+ models under a single interface. The backend runs on Supabase (PostgreSQL with pgvector for semantic search), payments through Stripe, and the entire stack deploys on Vercel. Agent Mode — which lets the AI write and execute real code in a sandboxed Node.js environment running in your browser — required significant engineering work on WebContainers and cross-origin isolation. Study Mode's page-locking and citation system was built specifically for students who need AI answers grounded in their actual textbooks.
The Philosophy Behind LumiChats
LumiChats is not trying to compete with Anthropic, OpenAI, or Google on model capability. It routes through their models. It is competing on the access layer: the pricing model, the student-specific features, and the multi-model workflow that lets users switch between the world's best AI systems without paying for multiple subscriptions. The founding insight was simple: premium AI should be accessible to students on student budgets. The ₹69/day price point was chosen because it is roughly the cost of a coffee or a cinema ticket — a discretionary spend that feels proportionate to a student's budget on a day when AI genuinely helps them.
Pro Tip: LumiChats is open source on GitHub under the account adityajhakumar. The README documents the full architecture including the RAG pipeline, WebContainer setup, Supabase schema, and payment flow — a reference implementation for developers building similar platforms.