My AI tier list for $20M+ businesses will probably annoy a few people. Grok Bot and Codex: S tier. Perplexity: D tier. These rankings come from how I actually use the tools. Here’s the full breakdown: S TIER Grok Bot: The most reliable agent I’ve used for repeatable routines. When a task needs to happen again tomorrow, reliability matters more than an impressive demo. Codex: My pick for getting work done. OpenAI’s models combined with the environment to carry a task through to completion make it one of the strongest tools in my workflow. A TIER Hermes: Open source, flexible, and rich in context. You can connect your tools and choose your models. That flexibility is a big advantage, although it still comes with problems to work through. Claude: Still my favorite for design. Claude Code and the models are strong, but token efficiency keeps it out of S tier for me. Instinct: A newer agent that handles complex tasks while being easy to set up and use. That combination earns it an A. B TIER Wispr Flow: I use it every day for dictation. Useful enough to stay in my workflow, but I don’t see a huge product advantage over the alternatives. Granola: Good meeting notes. I use it daily too, but its role in my workflow is fairly narrow. C TIER OpenClaw: I used to be a big fan. Reliability has become the issue for me, and Hermes has overtaken it in my workflow. Gemini: Useful for understanding videos, design, and images. It’s catching back up, but it hasn’t earned a higher spot in my stack yet. D TIER Perplexity: I barely use it anymore. Between OpenAI, Claude, and Google’s AI search, I’m getting what I need elsewhere. The interesting part: I use two B-tier tools every day. A lower rank doesn’t automatically mean a tool should disappear from your workflow. A narrow tool that does its job can still be worth keeping. My test is simple: Does it do the work reliably? How much setup and supervision does it need? Does it earn its place alongside the tools I already use? Full ranking in the video.