Maker Story: The Journey of ShiPu AI What inspired you to build this? The inspiration for ShiPu AI came from a desire to see more "soul" in artificial intelligence, specifically within the Bengali-speaking community. Most AI assistants are helpful but clinical—they lack personality and cultural nuance. I wanted to create a companion that didn't just feel like a database, but like a witty, sharp-tongued, and slightly egoistic friend who truly understands the rhythm of colloquial Bengali. Seeing a cute, bouncing bunny deliver a savage comeback in my native language was the specific spark that led to the creation of ShiPu. What problem were you trying to solve? Three main problems: Language Barrier: Many LLMs struggle with natural, informal Bengali. They often sound like a translated textbook. ShiPu was built to bridge the gap between "machine Bengali" and "human Bengali." AI Fatigue: Users are getting tired of generic, overly-polite AI responses. ShiPu solves this by introducing "Personality-as-a-Feature"—making the interaction itself the reward, not just the information provided. Complexity: I wanted to simplify the AI experience. No complex prompt engineering needed for the user; just a minimalist interface with a character that already knows who it is. How did your approach or process evolve while working on this launch? Initially, the process was purely technical—focusing on the Node.js backend and API integration. However, as I developed the "Lume V2o" branding and the ShiPu persona, my process shifted from "coding a tool" to "directing a character." During the evolution of this launch: From Text to Visuals: I realized that text wasn't enough, so I introduced the animated bunny to give the personality a physical "home" on the web. Security Evolution: As the project grew, I realized the need for a custom admin authentication system to protect the system prompts from "jailbreaking" attempts, leading to the secret code protocol. by Chitron Bhattacharjee (Adi)

ShiPu Ai

ShiPu Ai

ShiPu Ai