Built by @XForge0

I built CleanedIn because I kept opening LinkedIn with good intentions and closing it twenty minutes later having learned nothing. The feed had turned into something I didn't recognise — endless ChatGPT posts dressed up as insights, polls engineered for engagement, humble brags stacked six deep. Real posts from people I actually follow were buried somewhere underneath.

I looked for an extension that would fix it. There were a few options — things that blocked the feed entirely or filtered by keyword. Neither was what I wanted. I wanted to keep using LinkedIn. I just wanted to see less noise and more signal.

So I built it myself, mostly on evenings and weekends, mostly by crashing into problems and finding my way around them. The first version was embarrassingly rough. It got better over a few months of actually using it every day and fixing what annoyed me.

How it works

Each post gets sent to a server-side classifier — I'm using Claude Haiku, which is fast and cheap enough to run per-post without the cost getting out of hand. The classifier checks whether the post matches any of your active filters: AI-generated content, engagement bait, humble brags, and about seven others. If it matches, the post gets collapsed. You can still see it if you want. Nothing is deleted.

There's also a Spotlight feature on the Pro plan that does the opposite — it highlights posts that look genuinely worth reading with a blue left border. That part took longer to get right. It's easier to detect what's bad than to reliably identify what's good.

What I'm working on

The classifier gets better with time as I see where it's wrong. There are still edge cases — posts that are borderline, or genuinely personal content that looks like engagement bait but isn't. I tune the thresholds when I notice patterns.

Longer term I want to build something for other feeds with the same problem. LinkedIn was the obvious starting point because it's where AI-generated content landed hardest first, but the pattern isn't unique to LinkedIn.

If you have feedback, a bug, or a feature you'd like — find me on X or email getcleanedin@gmail.com. I read everything.