The author of the 'Iichko' channel built a working IT platform from scratch in 39 days without a technical background. During development, AI generated more than 35 million code tokens and processed 7.5 billion context tokens, which allowed automating the creation of architecture and interface.
What Happened
Using modern AI coding agents, a full-fledged IT platform was created, including 25 interface screens, an architecture of 27 database tables, and 81 software components. During the project, more than 62,000 lines of code were written and 1,463 tests were conducted.
Context
This case demonstrates the capabilities of advanced RAG systems and agentic cycles with a long context window. Using LLMs as the main design tool allows shifting the focus from writing syntax to high-level architectural thinking.
Why This Matters for the Industry
The case confirms a radical reduction in the barrier to entry for software development and the growth of the trend of non-technical founders who can quickly create complex MVPs. This leads to the transformation of the Junior Developer role into an Agent Orchestrator and the emergence of a new wave of micro-SaaS created by individuals.
Why This Matters for Users
For readers, this means that knowledge of programming language syntax is becoming less critical than the ability to manage AI agents. Technical expertise is no longer a blocking factor in the early stages of implementing business hypotheses.
What Is Still Unknown / Limitations
There is a gap in assessing applicability: if product roles see success in this, Enterprise architects point to the risks of technical debt, security, and IP compliance. The presented token metrics are more indicators of intensity than technical quality benchmarks.
Sources
Author
Look at AI, editorial team