Claude vs Granola

Claude is an AI assistant built by Anthropic, designed for professionals and operators who need more than a generic AI response. It tends to follow complex instructions more completely, sounds less generic, and holds onto more context when you give it a lot to work with. There's also Claude Code for developers and Claude Cowork, a desktop agent that works with files on your machine.

I'll be honest, I resisted trying Claude for a while. I'd been using ChatGPT for two years and it knows me and my writing style incredibly well. Every time I tried Claude before really committing to it, I was disappointed because it just lacked that context, and I'd walk away thinking "I don't get what everyone's talking about!" But the hype kept coming from people I respected, so I couldn't shake the feeling that I was missing something.

What changed it for me was exporting all of my writing instructions from ChatGPT and loading them into Claude as a skill. After that, the experience improved dramatically. Claude started asking better questions and the writing output got a lot closer to what I actually wanted. Now I find myself reaching for Claude specifically when I'm doing deeper writing work. It pauses to ask clarifying questions more than ChatGPT does, and it genuinely seems to honor the context I've given it rather than just acknowledging it and moving on.

Is that because Claude is naturally more thoughtful, or because my prompt told it to ask more questions? Honestly, hard to say. Without the skill loaded in, my earlier experience was nowhere near as impressive as the internet made it sound. So if you're going to try Claude and you already have a strong writing brief (or any other brief) somewhere, move it over before you form an opinion.

Where it still falls short is the surrounding product experience, the Chrome extension, the cross-device memory in Cowork. Those feel like areas where ChatGPT has just gone a lot deeper on the day-to-day user experience, and you notice the gap when you hit them. For now I'm keeping both, especially because I use ChatGPT Atlas for browser work.

• Automatically references your skills and writing briefs without being told to
• Longer context window means it holds onto more when you give it a lot of input
• Claude Code is loved by engineers

• Chrome extension has no memory, close it accidentally and everything is gone
• Cowork context doesn't persist across devices
• Usage limits are annoying, we hit them fast doing deep research on the standard tier
• Cowork is less useful if your whole workflow lives in web apps

Claude gives more complete answers without turning into an endless refinement loop. ChatGPT has a habit of ending every response with "want me to improve this?" and you end up going in circles. Claude just answers, and the conversation tapers naturally unless you push it further.

Claude automatically references your writing brief or skill in every relevant chat without you asking. With ChatGPT you have to actively add chats to a project to get that same behavior. It's a small thing that ends up mattering a lot when you're opening new chats constantly throughout the day.

Cowork is Claude's desktop agent, it can take actions across your computer, work through tasks, and integrate with your tools. If you work heavily out of local folders and project files, you'll probably get a lot out of it. For me, my whole workflow lives in apps like Superhuman Mail, Linear, and Granola, so it hasn't been life-changing.

• Free: Limited access, good for testing the product before committing.
• Pro: $20/month. This is the tier most founders and operators will want. Gives you access to the full model and the features that actually matter for daily work.
• Max: $100/month. Higher usage limits and more capacity for longer, more intensive sessions. Worth considering if you're hitting limits regularly.
• Team: $30/month per user (minimum 5 users). Adds collaboration features and admin controls for teams.
• Enterprise: Custom pricing. Expanded security, compliance, and deployment options for larger organizations.

Granola is a meeting recorder tool that doesn't require those annoying bots to join. In-fact the people you're on the call with don't even have to know you're using it.

Interestingly enough, this isn't any different than just having a note taker on the call, because they don't allow you to playback the audio. What you can do is ask Granola's AI Chatbot questions about the call, or any calls you've had across the company.

You can even ask Granola questions mid-call, which uses full context of the call up to that point, or can even come up with contextual questions to ask to help the conversation flowing. It's really as if you have an active note taking assistant on the call alongside you, helping you keep the conversation flowing, and researching further context details when needed.

We make sure to always have Granola recording our meetings (even in-person team meetings and brainstorming meetings), even if we have one of the best AI note taker bots joining the call as well. It's just a more lightweight solution that can't be kicked from calls and is less invasive.