8 AI Coding Personalities, Explained

The "AI Coding Personality Test" uses three independent binary dimensions to break down how developers work with AI into 2³ = 8 types. Below is a deep dive on each one — not just jokes, but targeted tool tips and growth directions too.

🤖 The AI Pope DVT-9000

—— Founder of the Vibe Coding faith — AI is basically my mom

You made AI the driver a long time ago and demoted yourself to co-pilot slash product manager. A requirement barely takes shape in your head before your hands are already typing the prompt. For you, coding isn't typing line by line anymore — it's a loop of "describe → generate → tweak." You love the flow state, and honestly your output speed is wild.

You'll Probably Relate

• Most of the code you ship in a day is AI-generated; you just steer the direction
• When you hit an error, your first move is to paste the whole stack trace back to the AI, not read it line by line yourself
• You genuinely can't remember the last time you hand-wrote a full module from an empty file

Tools & Workflows That Fit You

• Lean on Agent mode (Claude Code, Cursor Agent) and let the AI take tasks end to end
• Write your team conventions and directory layout into CLAUDE.md / .cursorrules — it noticeably cuts down how often the AI goes off the rails
• Build the habit of "AI proposes a plan, you review and approve, then it codes" — far more controllable than letting it edit straight away

🌱 My advice: Speed is your superpower, but draw yourself one line in the sand: you should at least be able to read and hand-edit the core paths. The day the AI goes down or sends you the wrong way, you shouldn't be dead in the water.

😰 The Anxious Vibe Coder DVS-404

—— The tortured artist who uses AI and curses it the whole time

You can't live without AI, but you never fully trust it either. Every time it spits out code, you grill it in your head — "is this actually right?" — and then use it anyway, doubting all the while. Your dev workflow is one tangled loop: use AI → doubt AI → verify AI → use it anyway.

You'll Probably Relate

• After you copy the AI's code you can't help re-reading it twice, terrified there's a trap buried somewhere
• You enjoy the speedup while quietly spiraling over "am I way too dependent on this thing?"
• You'll ask the AI the same question several times in different wordings until the answer finally puts your mind at ease

Tools & Workflows That Fit You

• Turn your doubt into process: have the AI write the tests alongside the code, and let the tests replace your endless manual double-checking
• For critical logic, ask it to explain "why it's written this way" — way more efficient than just staring at the code
• Set yourself a rule: ask the same question twice at most, then let the runtime results do the verifying

🌱 My advice: Your wariness is a strength — just don't let it turn into a spiral. Cash it out as automated checks (tests, type checking, lint) and a lot of the anxiety melts away.

📐 The Disciplined AI Engineer DRT-001

—— Trains the AI to be as buttoned-up as a direct report

You're a heavy AI user, but you never use it raw. You give it specs, context, boundaries — the AI is your senior assistant and you're the PM calling the shots. Your prompts read like short essays, but the output quality is a clear notch above too.

You'll Probably Relate

• Before touching anything, you write out the requirements, constraints, and acceptance criteria, then hand it to the AI
• You proactively feed it project background and the relevant files instead of tossing one sentence and demanding a result
• Sometimes you spend longer writing the prompt and assembling context than reading the code

Tools & Workflows That Fit You

• Use project-level context files (CLAUDE.md / rules) to bank long-term conventions so you're not re-explaining them every time
• Break complex tasks into small steps and confirm each one, rather than having the AI rewrite a huge chunk at once
• Commit often with git so every AI change has a clean rollback point

🌱 My advice: You've already turned the AI into a well-managed report — keep it up. The one thing to watch is over-engineering your prompts: for some small stuff, just doing it yourself beats writing half a page of instructions.

🔍 The Senior Prompt Engineer DRS-777

—— Uses AI hard, but reviews every single line three times

You lean heavily on AI for speed while staying on high alert. You have it write the code, the tests, the docs — then you go over all of it with the eyes of someone reviewing an intern. To you, AI is smart but unreliable: worth using, but it has to be verified.

You'll Probably Relate

• By default you treat the AI's output as "guilty until proven innocent" — you're not at ease until you've verified it yourself
• You habitually have a second AI or tool review the first AI's output
• You chase efficiency, yet occasionally wonder "is all this re-reviewing actually slowing me down?"

Tools & Workflows That Fit You

• Build a three-stage pipeline of "generate → automated checks → human spot-check," and save your human attention for the high-risk parts
• Review AI changes as a diff — focus on what changed instead of reading everything top to bottom
• Have two AIs review each other (one writes, one pokes holes) — it surfaces plenty of issues a single pass would miss

🌱 My advice: You're the team's anchor. Bake your verification instincts into a checklist or CI rules so the whole team benefits, instead of you being the only gatekeeper.

⌨️ The Tab-Key Warrior IVT-Tab

—— I don't depend on AI — I just can't live without the Tab key

You barely use chat-style AI, but you've worn the Tab key on Cursor / Copilot to a polish. You insist "I mostly write the code myself," it's just "letting it autocomplete a bit." The workflow looks independent, but without completion your typing speed would probably drop by half.

You'll Probably Relate

• You almost never open the chat window, but you can't go a second without inline completion
• You love the rush of "it nailed my next line" while refusing to admit that's dependence
• The moment completion glitches or the network lags, your whole flow falls apart

Tools & Workflows That Fit You

• Inline completion is best for boilerplate and repetitive structure — save your brainpower for the parts that actually need design
• Try chat-style AI now and then for whole-scale refactors that completion can't reach — it might open a new world for you
• Get fluent with the accept / reject shortcuts for completion — it makes the flow even smoother

🌱 My advice: You've kept your independent thinking, which is great. But don't write off chat-style AI entirely — for cross-file refactors and writing tests, it can do a lot more for you than completion.

🙄 The Stubborn Lone Wolf IVS-Solo

—— Uses AI occasionally, would never admit to depending on it

Your mouth says "AI's nothing special," but your body is honest — the moment you're stuck, you sneak off to ask it anyway. You believe in coding by feel, while staying skeptical of AI. Your commit history is dotted with little traces of "ugh, fine, I used it again."

You'll Probably Relate

• You trash-talk AI in public, then in private it's the first place you go the moment a hard problem hits
• After using it you'd rather not admit it — "eh, I could've handled this much myself"
• You have an instinctive distrust of AI-written code, but you also can't be bothered to write everything from scratch every time

Tools & Workflows That Fit You

• Instead of agonizing over "should I use it," just draw a clear line: what you write yourself, what you happily hand to the AI
• Treat AI as a docs / API lookup accelerator — low barrier to entry and easy on the "pride"
• Try logging how many times in a week the AI actually helped you — the data might change your mind

🌱 My advice: Being stubborn is fine, just don't let it cost you. Admitting a tool is useful isn't a loss — spend the time you save on the things actually worth fighting over.

🛠️ The Modern Old-School Dev IRT-Pro

—— Writes it carefully yourself, and trusts AI when it counts

You're the "do the heavy lifting yourself, AI as backup" type. You'd never outsource the architecture or core logic, but the grunt work — scaffolding, unit tests, docs — you're glad to hand to AI. Your relationship with AI is healthy, sustainable, and has clear boundaries.

You'll Probably Relate

• You insist on hand-writing the core modules, because you want to own every line of them
• Repetitive work and boilerplate you cheerfully toss to the AI without a second thought
• Your mindset around AI is steady — neither worship nor rejection, just "is it useful this time or not"

Tools & Workflows That Fit You

• Let the AI do the "boring but necessary" work: generating test cases, filling in doc comments, writing migration scripts
• For exploratory tasks, let the AI prototype first, then rewrite the critical parts on top of it yourself
• Write down the architecture conventions in your head and feed them to the AI — it'll fit your style much better as a backup

🌱 My advice: Your mindset is the model most people should be learning from. You can afford to be a bit bolder: some of the parts you're used to hand-writing, the AI can probably get 80% of the way there — try ceding more of the boundary.

⚒️ The Artisan Hand-Coder IRS-1989

—— "AI writes garbage — I'll do it myself"

You firmly believe code has to be hand-typed line by line; in your eyes AI is all show and no substance. Your vim config has been handed down three generations, and you use your dotfiles more than your ID. Watching coworkers vibe coding, your inner monologue is "these people are gonna pay for this sooner or later."

You'll Probably Relate

• You refuse to turn on any AI completion in your editor — you figure it just makes people dumber
• You have an instinctive disdain for AI-generated code, and you'd rather write it slowly yourself
• Your fundamentals are rock-solid; you're the person on the team who "understands all the low-level stuff"

Tools & Workflows That Fit You

• You don't have to buy in wholesale, but you can treat AI as "a search engine that chats back" — looking up errors, looking up API usage
• Even just trying it in low-risk spots like writing tests or comments can save you a fair bit of time
• Reading AI-generated code can sometimes help you spot blind spots in your own thinking

🌱 My advice: Your solid fundamentals are a rare commodity. But the AI era is redefining efficiency — leaning on it a little doesn't mean abandoning the craft. Apply the craft where it matters, let the tool carry the repetitive work, and you'll save yourself a lot of effort.

So, which one are you?

Out of the 8 types above, there's always one that makes you go "okay, that's literally me." Take two minutes and find out → — 12 random questions, and at the end you get a personality card worth sharing.