# Python Is Not Enough: Why Pythonistas Love Rust (Podcast) > Source: > Published: 2026-06-25 00:00:00+00:00 # Python Is Not Enough: Why Pythonistas Love Rust (Podcast) *Learning Rust? I co-run a 6-week Python to Rust cohort where you build a performant JSON parser with PyO3 bindings.* I joined [Bas Steins](https://www.linkedin.com/in/bascodes/) and [Michal Martinka](https://www.linkedin.com/in/michal-martinka/) on their complexity.fm show to talk about why Pythonistas are picking up Rust, what AI really does to how we learn, and why vibe coding is a myth. The conversation ran for over an hour because there was a lot to unpack. Watch the full episode on YouTube: ## Highlights from the episode - Why Pythonistas reach for Rust - A lot of tooling is created and rewritten in Rust (Ruff, uv, Polars, Pydantic core). - Rust's strictness offers better guardrails for non-deterministic agents. Open Python vs closed Rust (mutability opt-in, single owner, compiler rejects skipped cases). - Fighting the borrow checker changes how you write Python. - Related: [Learning Rust made me a better developer](/blog/rust-made-me-a-better-python-developer)and[The Rust compiler as an AI coding agent guardrail](/blog/rust-compiler-ai-agent-guardrail) - You can get far with Python's type checkers - We discuss how for business logic and when performance is not essential, Python + type checkers gets you far. - What Rust still gives you though is exhaustive enum matching enforced by the language. - The tooling got a lot better: [Modern Python tooling: uv, ruff and ty](/blog/modern-python-tooling-uv-ruff-ty) - Vibe coding is a slot machine - Real dopamine effect, working with multiple agents has a lot of context switching, which is bad for productivity (and your brain). - The real speed up number: not 10x, more like ~1.2-2x. You get a prototype in an hour, but then the iteration/judgment/cleanup often takes days. - The rubber-stamping risk: code that looks plausible, but you miss the 50 lines duplicated three times, because you didn't feel the pain of producing the code. - Timeless coaching that outlives the hype - All project-based, it's the best way to learn holistically, but noticing a shift from greenfield (2020) to more brownfield and AI hardening these days. - Ryan's Payroll SaaS case study: tools changed, TDD/guardrails mindset didn't and stemmed from the coaching years ago ("don't give the fish, teach how to fish"). More on this: [AI coding tools fundamentals case study](/blog/ai-coding-tools-fundamentals-case-study/). ## Where to go from here Michal ends on a positive note: I think the world is actually starting to heal. I think there will be a lot of adjustment with all this AI hype. And it's something I am coming back to as well: yes, we have a powerful set of new tools, but they are only as good as the knowledge and judgment you bring to them. The hype is deflating into something more honest, and the tools aren't going anywhere, but neither is the need for engineers with deep skills. If you want to further develop this skill set, check out [how I help engineers level up](/coaching/). Want Rust to click beyond syntax? Build a JSON parser from scratch, wire it into Python with PyO3, and benchmark it against CPython and other JSON libraries. Six weeks of practical Python to Rust engineering with weekly PR reviews and support by experienced Rust and Python engineers, not lectures. [Join the next Python to Rust cohort →](https://scriptertorust.com)