Make AI Score Itself: Shipping Products Without Writing Code

Zero Lines of Code, Multiple Shipped Products

SCAILE co-founder Federico De Ponte has never written a single line of code by hand. Yet he has shipped multiple functional products - from web scrapers to AI assistants to design tools - all built entirely through conversations with AI coding agents like Claude Code.

In a recent breakdown of his workflow, Federico shared the key insight that makes this possible: the trick is not just telling AI what to build, but making AI evaluate whether what it built actually works.

The Self-Scoring Method

The approach hinges on a feedback loop. Rather than reviewing generated code line by line (which would require the programming knowledge Federico freely admits he lacks), he instructs the AI to test and score its own output against a set of criteria.

The workflow looks like this:

• Describe the desired outcome in plain language, focusing on what the product should do, not how it should be implemented
• Ask the AI to build it, providing context about the target users and use cases
• Have the AI evaluate its own output against specific success criteria - does it handle edge cases, is the interface intuitive, does it perform under load
• Iterate based on the AI's self-assessment, refining the prompt and constraints until the product meets the bar

This self-evaluation step is what separates productive AI-assisted development from the frustrating cycle of generating broken code and not knowing how to fix it.

Building in Public at SCAILE

Federico's approach is not just a personal experiment - it reflects SCAILE's broader philosophy. The company operates as an AI-native organization, using AI tools at every level of the business. From marketing content to internal tooling to client deliverables, AI is embedded in the workflow rather than bolted on as an afterthought.

Several of the tools Federico has built and open-sourced - from web scrapers to design generators - were all created using this same method. Each one started as a conversation with an AI coding agent and was refined through iterative self-scoring.

What This Means for Builders

Federico's message to other non-technical founders is direct: the barrier to building software has effectively collapsed. The remaining skill is knowing how to describe what you want clearly and how to set up feedback loops that catch problems early. Those are communication skills, not engineering skills.