Coding as a data person is wild these days. You’ve got SQL, Python, statistics, visualisations, pipelines, some weird internal API, dbt, Airflow, DevOps, a side of git conflicts, all before your first coffee kicks in. But, now there’s this new wave of tools that promise to make it all easier. AI editors (also known as autocomplete on steroids). The most popular one is Cursor, a code editor that aims to build AI directly into your workflow. If your first reaction is “ehhh I don’t know about that,” you’re not alone. A lot of folks in data are skeptical about AI writing or even helping with code. And honestly? That skepticism makes total sense. But here’s the thing: when used right, Cursor is a powerful tool. Think less “robot dev writing my code” and more “super-fast assistant who doesn’t mind writing documentation or figuring out why my script won’t run.” Let’s dig in.
So, what is Cursor?
Cursor looks and feels like VS Code, but it has a built-in AI assistant that actually knows what’s going on in your project through the context you give it. You can highlight some code and ask it to refactor it. You can type a comment like “write a function that fetches yesterday’s data from BigQuery” and it’ll give it a shot. You can ask what a certain block of code does, and it’ll explain it in plain language. The key difference is: it’s not just guessing in the dark. Cursor reads your files, and uses the same liniter and documentation as you. So when it makes suggestions, they’re often surprisingly relevant.
Why the skepticism is fair
A lot of data folks are wary of this stuff, and for good reason. You’ve probably seen those AI-generated code snippets that look right… until they very much aren’t. Or the “tab-tab-tab-run” approach where you get a working solution, but have no idea how or why it works. Not ideal when you’re on call for that job next month.
There’s also just pride in writing clean, thoughtful code yourself. We didn’t spend all that time learning how to structure queries or debug DAGs just to hand it off to an LLM, right? Totally fair. We should be critical. AI still messes up. It still hallucinates. It still sometimes gives you answers with scary confidence and zero correctness. But here’s the perspective shift that helped me.
In 1966, teachers were mad about calculators
Seriously, there were actual protests when calculators were introduced in schools. The fear was that students would stop learning how to do math by hand. That they’d get lazy. That math education would collapse.
Fast forward a few decades and no one’s doing long division on a whiteboard during a meeting. The calculator didn’t kill math, it just freed people from doing tedious stuff over and over, so they could focus on understanding, modeling, and solving more interesting problems.
I think AI coding tools are at a similar moment. Used badly, they absolutely can be a crutch. But used well, they’re just… a better calculator. A tool that lets you spend less time writing boilerplate and more time actually thinking.
So how do you use Cursor without becoming a tab zombie?
Here’s what’s worked for me: treat Cursor like a very smart intern. It’s great at doing tedious things, suggesting unexpectedly brilliant ideas, and writing first drafts. But as the old saying goes: trust is good, control is better. You still need to review, guide, and drive the vision.
So, what can you use it for?
-
When you clone a messy internal project with no documentation, ask Cursor to generate a first draft for a new README based on the code. This provides a solid starting point for your own understanding, rather than figuring it out from scratch.
-
If you’re setting up tests and dreading repetitive boilerplate, Cursor can handle the scaffolding, letting you focus on the parts that require your attention.
-
When dealing with spaghetti code, highlight it and ⌘+k Cursor for a refactor. Even if it’s not perfect, it often helps get you moving and spark some ideas.
-
If you’re unsure how to organize a script, structure a repo, or break up logic into modules, ask Cursor for different approaches and pick the one that makes the most sense.
The point isn’t to get it to write your logic for you, the point is to get it to handle the 30% of stuff you already know how to do, but don’t really want to. This isn’t about turning your brain off. It’s about turning the busywork down.
Final thoughts
Cursor isn’t magic. Sometimes it’s a game-changer, sometimes it throws utter nonsense your way. But when it works, it’s like having a coding buddy who’s already combed through every Stack Overflow thread out there.
If you spent a lot of time with code, tools like this are worth a real try. Stay skeptical, stay curious. You don’t need to go full “AI workflow wizard” to get value out of it. Use it where it helps and skip it where it doesn’t. Also be aware that the tooling and underlaying models change constantly, what doesn’t work this week might work the next.
In the end, I believe that the best data folks won’t be the ones who’ve memorized every Pandas method. They’ll be the ones who combine experience with smart tooling to get to better results, faster. Cursor won’t make you a great coder. But if you already are one, it might just help you move like it.
Hi, I’m Bastiaan 👋🏼 Founder of datalyft, a small data agency helping companies transform their raw data into real value. I write about the Modern Data Workflow, where I explore tools & processes to supercharge your data capabilities. Follow me for more!