If you want something done right, automate it.

Rust is condemned to succeed!

• rust’s globla code base is tiny compared to almost everthing else, case in point C++ and C

• but it is memory safe

• and equally performant

• and at least as expressive

• it has super high quality compiler feedback

• this makes it a perfect match for coding agents which loop until the succeed

• the high quality compiler feedback is key, agents can converge to a good solution

• the existing traing code base is small, but relatively hight quality (AWS, Microsoft, etc). Search and provide examples hire

• code bases like the linux kernel will make progress, despite the fact that existing maintainers are reluctant. Time will solve this issue.

• find other reasons why rust is in the sweet spot for agentic coding and therefore will dominate !!!

• rust is the most loved language 7 years in a row, but this is irrelevant

• Programming agents need strong typing and compiler-enforced correctness

• fullstack capabilities: not talking about fullstack web development, but code for embedded microcontrollers, mobile devices, big servers, satellites and everything in between with a single programming language

helsing

Top Rust users like helsing are using Claude collaboration to create PRs: Pull request search results · GitHub

Why not Golang?

• Go’s type system and compiler are not sophisticated enough for AI agents. There are just too many cases of “it compiles but crash in production”. Find examples
• There are just too many footguns that are not enforced by the compiler: nil pointer null pointer dereference, implicit default values, where if you (or the model) forget to populate a field of a struct at creation it may lead to unexpected behavior, harmful enums and more…

Source

Read this: https://kerkour.com/rust-agentic-coding and use its ideas and findings for this article

Prompts

Test Hypothisis: High Quality Code

• You are a expert in software development and programming languages.
• You should verify or falsify the hypothesis: Existing Rust code base is small, but high quality overall.
• Large Language Models need huge data sources for their training, so computer languages with a huge code base are supposed to have an advanvtage in coding agents due to the vast training material.
• But: high quality data is more important than volume which incorporets also lots of bad practices and bug-prone programming, e.g. C++ due to its evolvment of time
• Rust does have small code bases
• Analyse the hypotheses that the code base of Rust is relatively high quality due to its youth and incorporation of many lessons learned from C++ and other languages
• high learning curve results in more senior programmers using Rust for serious projects (this is alos a hypothesis)
• collect evidence for and against the hypothesis and make an proefssional assessement. Provide examples.
• In the end it should be a convincing case that the existing training body of Rust for LLMs results in very good LLM mastering of Rust based on modern best practices. however do not hesitate to falsify this statement if the facts go in this direction.