LIVE: You NEED to be using OpenClaw and Claude Code together
TL;DR
Alex Finn’s core point is anti-tribalism: use Claude Code, OpenClaw, and Hermes together — he argues Claude Code is best for focused app building, while OpenClaw and Hermes shine as broader “advisor” agents with memory, file access, and cross-workflow support.
He calls the new Claude Code desktop the best vibe-coding tool he’s used, even better than the CLI — the reasons are project/session organization, customizable panes, yellow-dot approvals, and bypass-permission mode, which he says makes multitasking dramatically smoother.
Claude Code gets the coding seat because its context stays clean and app-specific — Finn says he builds Henry Intelligent Machines with Claude Code, while OpenClaw and Hermes help with planning, feedback, and operational support because they remember more about him and his work.
He pushes back hard on the idea that OpenClaw is “dead” — his argument is that truly dead tools die quietly, while OpenClaw is still shaping the market, with Anthropic’s recent routines and agent moves acting as clear responses to it.
Finn is unusually blunt about creator trust and AI sponsorships — he says he’s turned down a $30,000 Hostinger deal, repeatedly trashes undisclosed Higgsfield promotions, and frames trust as the most valuable asset an AI creator can build.
His broader message is that this AI window rewards speed, discomfort, and self-teaching — he repeatedly tells viewers to become “autodidactic,” try every major update fast, and use AI to build real products instead of chasing lazy schemes like agent-driven trading.
The Breakdown
The stream starts with a rant against AI tribalism
Finn opens by saying the whole “only OpenClaw,” “only Claude Code,” or “only Hermes” mindset is dumb. His thesis for the entire stream is that the winning workflow is mixed-tool: stop treating AI products like sports teams and start using whatever actually works.
Claude Code desktop is the “story of the week”
Once he gets past the usual live-stream chaos, he dives into the new Claude Code desktop and says it’s the best vibe-coding experience he’s had so far. He likes the project-based organization, multiple sessions per project, attention indicators like yellow dots, and the ability to pin preview/files/plan/task panes however you want — while also noting a few UI bugs when switching sessions.
Why he now prefers desktop over Claude Code CLI
Finn says he used to be a CLI-first person, but the desktop app changed his mind. He compares Anthropic’s move to OpenAI/Codex: Anthropic basically took the best parts of competitors’ app UX, skipped the “vague posting,” and shipped what he calls a full-on super app before OpenAI could fully land its own version.
The actual workflow: Claude Code builds, OpenClaw and Hermes advise
This is the clearest practical section of the stream. For serious consumer apps like his company Henry Intelligent Machines, he uses Claude Code for the actual coding because it has tighter context and is trained for that job; OpenClaw and Hermes sit beside it as “advisors,” helping with decisions, ideas, and broader context because they know more about him, his history, and his files.
OpenClaw isn’t dead — social hype just moved on
Finn addresses a question he says he keeps seeing online: has OpenClaw hype died? His answer is no — and he makes it in classic Alex Finn fashion: dead apps die in silence, while people are still loudly debating OpenClaw because it matters. He also points out that many recent Anthropic releases, especially routines, look like direct responses to the OpenClaw wave.
Where OpenClaw and Hermes differ — and where they don’t
He refuses to pick a clean winner between Hermes and OpenClaw, which is kind of the whole ethos of the stream. Hermes feels faster to him, OpenClaw’s memory has been more stable in his setup, both can break depending on how users configure them, and his favorite practical trick is simple: when one agent fails, have the other one fix it.
Creator trust, scams, and why he won’t play the influencer game
A huge chunk of the live is really about media ethics. Finn says the AI creator economy is full of people doing undisclosed shills, especially for Higgsfield, and he keeps coming back to the same point: sponsorships are fine if you genuinely use the product, but trust is worth more than a quick check. He contrasts that with turning down a $30,000 Hostinger offer and says the easiest way to ruin your reputation right now is to sell out.
The stream ends as a full-on motivational sermon about AI opportunity
The final stretch turns into a long, high-energy monologue about risk, urgency, and this being the biggest AI opportunity window in history. He tells viewers to become autodidacts, build things that scare them, stop waiting for permission, and use tools like Claude Code, OpenClaw, and Hermes to create value — not to run low-edge gimmicks like autonomous stock trading.