Corporatify - A Chrome extension for people who are tired of pretending

LinkedIn is a fever dream. It's a place where people get laid off and immediately post about it like they just completed a spiritual retreat. Where someone's six-month junior dev internship gets described as "spearheading cross-functional synergy initiatives." Where the most engaged content is always someone's deeply personal struggle repackaged as a productivity lesson, aaand then there's the seven-line gap before the call to action.

Theatrics. All actors (even though I hate to think that some actually aren't 💀). And most of us, at some point, have typed out exactly what we thought and then deleted it, because apparently "this entire thread is insufferable and it makes me want to leave the platform" is not considered a value-adding contribution.

So I built something to help with that. It's called Corporatify.

What It Is

Corporatify is an extension for Chromium based browsers that lives in the side panel. You type whatever you actually mean - unfiltered, uncensored, unhinged - and it spits back a professionally reworded version that you can actually post without consequence.

It connects to an AI backend of your choice: OpenAI, Anthropic, Groq, or a local model running through Ollama. You bring your own API key. There's no middleman, no account, no monthly subscription to a SaaS tool that charges you twenty dollars a month to wrap GPT in a slightly different trench coat.

You can also highlight any text on any page, right-click, and hit Corporatify this. The side panel opens and handles it automatically.

The Use Case Is Perfect, Actually

Yes, I know. Local AI models are, broadly speaking, kind of dumb. They summarize things confidently and incorrectly. Ask a small Ollama model to write you a business plan and you'll get something that reads like it was generated by a toaster that briefly attended a TED talk during lunch break on a Tuesday.

But here's the thing: for this specific task, small models are genuinely fine and the job is not hard. You're asking it to take "I did all the work and Dave got all the credit again and I'm about to lose it" and turn it into:

I wanted to flag that I played a central role in delivering this project and would appreciate it if that contribution were reflected in how it's communicated to stakeholders going forward.

A 6-7 (ayyy 🤲) billion parameter model can absolutely do this. It turns out that transforming impotent rage into corporate-speak is a task that maps almost perfectly onto what these models are actually good at: pattern-matching to a tone they've seen ten million times in their training data.

LinkedIn prose is, after all, one of the most formulaic things ever written.

Some Examples

Here's a small sampling of real inputs and what comes out the other side.

Input: this meeting could have been an email

Output: I wonder if this might be more efficiently handled asynchronously - happy to put together a brief written summary for everyone's review instead.

Input: nobody asked for your opinion but here you are again

Output: Thank you for sharing your perspective - it's always valuable to hear a range of viewpoints on topics like this.

Input: I hate it here

Output: I've been reflecting on some aspects of my current environment and I'm keen to explore ways to better align my day-to-day with my longer-term professional goals.

Input: fired

Output: Excited to share that I'm officially open to new opportunities! After an incredible run at my previous company, I'm ready for the next chapter.

That last one. I want you to sit with that for a moment.

Input: I did all the work and Dave got all the credit again and I'm about to lose it

Output: I wanted to flag that I played a central role in delivering this project and would appreciate it if that contribution were reflected in how it's communicated to stakeholders going forward.

Input: please stop messaging me I don't want this job

Output: Thank you so much for thinking of me for this opportunity. At this time I'm not actively exploring new roles, but I appreciate you reaching out and I'll keep your details on file.

This is perhaps the most impressive one. It takes a message that is essentially "leave me alone" and transforms it into something that sounds like the person is genuinely grateful for the interruption.

Running It Locally

If you'd rather not hand your unfiltered workplace grievances to OpenAI's servers - which, given the data collection and profiling is fair - you can run it entirely on your own machine with Ollama.

Install Ollama, pull a model, point the extension at http://localhost:11434/v1, leave the API key blank, and you're done. Nothing leaves your machine.

For model recommendations, since I know you're going to ask:

General use - actually good output:

• gpt-oss:20b - best local quality, needs a real GPU, worth it if you have one
• deepseek-r1:8b - solid balance of speed and quality, runs on most machines

Multilingual support - if you're writing in French, Spanish, German, or anything else and want it to stay in that language:

• translategemma:4b - lightweight, punches above its weight for multilingual rewrites
• aya-expanse:8b - better quality multilingual output, costs more VRAM

Pulling a local model with Ollama in CLI:

ollama pull gpt-oss:20b

Set the model name in the extension settings to match whatever you pulled, and you're off.

Note: to use a locally running Ollama instance you need to set a system/user or container (if you're using that) environment variable OLLAMA_ORIGINS=* and then restart your browser.

Site-Specific Prompts

One feature that's actually useful: you can set a different system prompt for each website. The extension detects what domain you're on and switches the prompt automatically.

Out of the box, LinkedIn gets its own prompt tuned to LinkedIn-speak. But you can add Slack, your email client, whatever internal tool your company insists on using - and give each one a different tone. Formal for email; Passive-aggressively polite for Slack; Vaguely inspirational for LinkedIn - the full suite of masks for modern professional life, now automated.

On AI Being Shoved Into Everything

We need more data centers Anakin, to cure cancer, right? 👀

Googling this is questionably SFW

We train bigger models and we deploy them everywhere. They run on hardware that consumes enough electricity to power a small country so that nobody ever has to think again (and I professionally specialize in this btw 😁) - optimizing the human out of the loop one token at a time.

But Corporatify is genuinely one of the few cases where this makes sense. It's not doing anything profound but rather something small and annoying - and it turns out that's exactly what these models are good at! It saves you thirty seconds translating your own feelings into a dialect you didn't choose and don't particularly like.

The AI doesn't care why you're angry at Dave 👊. Nobody cares about Dave anyway, but it just needs to know what angry people sound like when they're forced to be professional about it.

Getting It 🫰

The extension is on Chrome Web Store.
It's compatible with any Chromium browser - Edge, Brave, Opera, Chromium, Vivaldi, DuckDuckGo, etc.

Grab your API key from whichever provider you want to use - or just run Ollama locally - point the settings at it, and you're ready to say exactly what you mean without saying exactly what you mean.

Let's keep it civilized and restrain ourselves, so we don't end up on an NSA domestic terrorist list - Use Corporatify!

Live long and prosper. 🖖👽