Fourteen-Year-Old Developer Creates AI Tool Supercompress to Reduce Token Costs
Arjun Shah, a 14-year-old self-taught developer of Indian origin living in San Jose, California, has developed an artificial intelligence tool aimed at addressing one of the primary financial challenges of current-generation AI: the high cost of tokens. Shah, the founder of Supercompress, designed the tool to lower the growing expenses associated with AI tokens through calculated efficiency.
Whenever users interact with AI models, they consume tokens. The volume of tokens used increases with the amount of context provided, leading to rising operational costs. Shah explained that AI applications are forced to transmit massive amounts of context, the majority of which is redundant, yet companies must still pay to process all of that data, compounding expenses over time.
How Supercompress Reduces AI Token Usage
According to Shah, Supercompress is a neural network structured like a human brain with nodes and connections, which he describes as a framework for AI. The tool analyzes both the query and the context, identifies the critical information, and removes the redundant data before sending the input to the AI model.
"On average, it cuts by 65 per cent but keeps more than 98 per cent of critical content," Shah said. In a demonstration video, the tool compressed context by 97.5 percent, delivering the same answer as the original uncompressed input but at a faster speed and with lower token consumption. The creator believes the tool has the potential to save millions of dollars for companies globally. At the time of writing, Supercompress has approximately 150 users and is available as a plugin for coding agents.
A Non-Traditional Approach to Software Development
Shah began experimenting with coding at the age of seven using Python. Although he received some basic instruction at school, he was largely self-taught. However, he decided to stop traditional coding at the age of nine due to a dislike of memorizing syntax and technical details, though his interest in building remained.
The emergence of modern AI tools allowed Shah to continue building without focusing on technical coding. He began using the AI coding platform Cursor two years ago, using it to build an AI-powered plant identifier for a school science fair. He then spent a summer studying the mathematical and algebraic fundamentals of neural networks rather than learning how to write the code for them.
Y Combinator Application and Family Background
Shah has applied to Y Combinator, a prominent Silicon Valley startup accelerator that has previously backed companies like Airbnb and Reddit. If his application is accepted, Shah will receive $500,000 (approximately Rs 4.8 crore) in funding alongside access to the accelerator's network of founders. While the selection process is highly competitive, Y Combinator has previously funded teenage founders, including a 14-year-old and a 16-year-old approved by former president Sam Altman in 2017.
Shah's parents, who are now US citizens, originally grew up in Gujarat, India, before relocating to the United States. His mother has supported his development journey and previously served as the co-founder of his prior project, therooted.ai, which is a retrieval-based tool that references traditional texts like the Charaka Samhita to suggest remedies for everyday illnesses.
