Ruby's Hidden Advantage: Why AI Agents Code Faster in a Language Nobody Expected
Ruby's Hidden Advantage: Why AI Agents Code Faster in a Language Nobody Expected
When you claim that Ruby is the superior choice for AI-assisted development in 2024, you're essentially asking for trouble. The internet has strong opinions about which language matters, and Ruby isn't exactly trending on Hacker News anymore.
Yet someone just tested exactly that claim—and the results are surprisingly compelling.
The Experiment Nobody Asked For (But Needed)
Coolhand Labs recently ran a controlled experiment that had been nagging them for months. Instead of debating which language AI prefers, they decided to measure it directly. The task was straightforward: implement the same feature across Ruby, TypeScript, and Python client packages using Claude Code with Sonnet 4.6.
The workflow was rigorous:
- AI agents planned the solution
- Humans approved the plans unchanged
- A separate AI agent (Opus 4.7) performed code review
- PRs were created and fixed until CI passed
The results spoke louder than any LinkedIn post ever could.
The Numbers Don't Lie (Mostly)
Ruby crushed it. Tasks completed faster, token consumption stayed lean, and the workflow felt almost natural. TypeScript came in a solid second, though one particularly messy run skewed the results. Python... well, Python took a detour into a dark forest of inefficiency.
But here's what matters: when they dug into why Python struggled so dramatically, they discovered something far more interesting than just "Ruby is faster."
The Custom Instructions Plot Twist
The Python repository had been configured with custom instructions that required the AI to always run linting and tests before finishing a task. Ruby and TypeScript lacked this guardrail.
This detail is crucial. It's not that Python is inherently slower for AI agents—it's that how you prompt your AI changes everything. The custom instructions forced the agent into a more defensive, verification-heavy workflow. More steps. More tokens. More time.
This reveals something important about agentic coding: it's not just about the language syntax. It's about the feedback loops, the constraints, and the expectations you bake into your AI instructions.
What This Means for Your Stack
If you're building products on NameOcean's hosting infrastructure and considering which language to use for AI-assisted development, here's the practical takeaway:
Language choice matters less than workflow design.
That said, Ruby's advantages make sense when you consider its strengths:
- Concise syntax that's easier for LLMs to parse and generate
- Strong convention-over-configuration patterns that reduce decision trees
- Mature frameworks (Rails) that AI agents already understand well
- Fewer boilerplate requirements compared to TypeScript
TypeScript's performance gap is smaller than you'd expect, which suggests the ecosystem is catching up. Python's initial struggles highlight a real issue: its verbose structure and testing conventions can create longer token sequences.
The Real Question: Is This Actionable?
Before you rewrite your Python microservices in Ruby, take a breath. The experiment shows correlation, not necessarily causation about language superiority. What it does demonstrate is that:
- AI agents have measurable performance differences across languages
- Custom instructions and workflow constraints have outsized impact
- Token efficiency matters when you're running large batches of agentic tasks
- The tooling and frameworks around a language matter as much as the language itself
For teams running on our cloud hosting platform considering agentic development workflows, consider optimizing your prompts and instructions before switching languages. The token savings from better-designed constraints could dwarf any language-specific gains.
The Rage Bait Wasn't Wrong After All
Sometimes the most controversial claims deserve a closer look. This experiment proves that Ruby's resurgence in the age of AI isn't nostalgia—there's genuine technical substance behind it.
Whether Ruby becomes the preferred language for agentic coding remains to be seen. But we now know that the person who suggested it wasn't just stirring the pot. They were onto something.
The future of development isn't about picking the "best" language. It's about understanding which languages and workflows your AI agents work best with—and that's a conversation worth having.