Just under eight inches long and shaped to a sharp point, the Blackwing 602 might be the most heralded pencil in the world.

Incense cedar surrounds a firm graphite center. A flat metal cap clamps down over a thin pink eraser. It was a favorite of animators until it was discontinued. In 1998! That’s when a single pencil climbed to $40 or more on the secondary market. Another company eventually started reproducing them, but you can still find originals on eBay. Creators buy them up just to tap into that special something.

Disney animator Shamus Culhane, who led the original Snow White team, was buried with a Blackwing 602 in his hand.

That’s how much animators loved this pencil.

I had to get a few of the reproductions to see for myself:

If Shamus were still around, he could probably show me a thing or two with these things. How to extend the flat eraser with the built-in clip. How the weight sits in your hand on the thousandth frame of the day. Real tricks from someone who used the tool for decades.

But none of it would make me an animator.

I think about this a lot with developer tools. A great dev tool is still just a tool. The teams that win aren’t the ones with the shiniest API or the cleanest docs—they’re the ones who understand which use cases their tool is actually best for, and who can frame the product around those use cases in a way developers immediately recognize.

“I have logs scattered across three services and can’t trace a single request.”

“I need to test webhooks, but I can’t expose my local server to the internet.”

“Marketing wants full-text search across our content, and it’s all stored as Markdown now.”

That’s the language that makes a developer reach for a tool.

And now there’s a new wrinkle.

A lot of the time, the “developer” reaching for your tool is actually an LLM acting on a developer’s behalf. Which means the use cases you frame your product around aren’t just for humans anymore. They’re landing—or not missing—with the models that could recommend you.

The pencil doesn’t pick itself up. Something has to decide it’s the right one for the job. In animation, it was the Shamuses. In many dev workflows now, it’s Claude or Codex working from whatever impression they’ve formed of your product.

Which raises a pretty practical question: what impression have they formed?

​LLM Rank can tell you across 33 categories and counting, with the results recently updated every month. The leaderboards show which tools Claude, ChatGPT, and Gemini recommend.

It’s the closest thing I’ve seen to a feedback loop on how AI currently understands your product—and whether it’s pointing the right developers your way.

I also wrote up some lessons learned from watching these leaderboards.

How are you tracking when you show up?