Why I'm All In on Claude
In more than three decades around enterprise tech, nothing resembles the current AI wave. Its pace of change has far surpassed the other shiny new toys I've seen in my lifetime: the early web, social networks, and mobile apps.1
Best-of-breed tools pop up every day. Dedicated and popular newsletters like WonderTools cover dozens of them every month.
There's no shortage of powerful large language models powering intuitive AI chatbots. Increasingly, though, I've been gravitating towards one. In this post, I'll explain why.
Innovation
Last July, I wrote about the genAI tool that was blowing my mind. Anthropic didn't only invent the concept of model context protocols; the company released it as an open-source project. To paraphrase Gandhi, actions express priorities. It's obvious that Anthropic doesn't just value making a buck.
Utility
Custom GPTs aren't exactly news. Google recently dropped Gems—its version of bespoke AI apps. For its part, Claude allows customers to create custom Skills. In a word, they are sick. Just in the past week, I've created ten incredibly useful ones for my publishing outfit alone. Needless to say, many (most?) of these are far too esoteric for proper software developers to build them.
And then there's my LinkedIn profile evaluator:
Phil Simon
It takes zero technical skill to create and publish niche Claude Artifacts, such as spam detectors. (Citizen developers rejoice.)
The adage “Garbage in, garbage out” hasn't changed one bit.
I'll continue geeking out on Skills and Artifacts in 2026. You can take the kid out of Carnegie Mellon…
Effectiveness
No AI tool bats 1.000. The fleas come with the dog in the form of hallucinations. Despite my 1,200 or so chats with Claude in the past year or so, though, it has erred very few times. (I'll put that number at about two percent.) Of course, that hit rate may stem from the quality of my prompts. The adage “Garbage in, garbage out” hasn't changed one bit.
Brass tacks: If Claude erred 10 percent of the time, I'd find another AI chatbot.
Safety and Responsibility
Some companies pay lip service to the very real dangers AI poses to society, employment, truth, and politics. (You know, little things.) By contrast, Anthropic seems to take these existential issues far more seriously than OpenAI, X, and other behemoths. Exhibit A: Critical public service announcements such as these.
Forget just releasing professional videos. The company worked closely with The Wall Street Journal to see how unsupervised AI would do in the real world.
Design
I won't mince words here: Claude is a freaking joy to use. Its minimalist design aesthetic reminds me of Notion and Apple. (To be fair, Gemini, Copilot, Perplexity, and their ilk sport equally spartan user interfaces.)
You may read my rationales and guffaw. Let's say that the major AI chatbots essentially offer feature parity. That is, most product features are nearly identical (or will be soon). Does any attribute really distinguish Anthropic from its competitors?
The answer is yes.
Ethics, Hypocrisy, and the Right Side of History
Many AI companies are attempting to make the case that their blatant theft of pirated books is perfectly legal. They are bastardizing the very notion of the Fair Use Doctrine. Of course, they sing a far different tune when tech companies steal their intellectual property. (If this sounds hypocritical, trust your judgment.)
Not Anthropic.
Its settlement with authors and publishers is nothing short of trailblazing. At some point in the near future, Anthropic will remit me thousands of dollars. The company used some of my books to train its models—as well as those of many others.
Exhibit B: The Trump administration has proposed a dangerous laissez-faire approach to AI regulation. Rather than fall in line, Anthropic exes are fighting for more responsible government oversight. That willingness puts the company on the right side of history, and it should mean something to anyone with a modicum of common sense.
Phil Simon
The Bottom Line
I'm not completely monogamous. I still play with other AI tools—and you should, too. Google's image-generation capability is amazing. (See the featured image above.) NotionAI is certainly helpful, especially when it's not forgetting what its underlying product can do. For the foreseeable future, though, Claude is my AI go-to. I'd rather give my money to ethical companies.
Footnotes
-
No, the 900-user metaverse doesn't count. Not even close. Can someone say all hat, no cattle? Twenty bucks says that Zuck would love to take a mulligan on his ill-advised 2021 Meta rebrand.
Member discussion