Codex vs Claude for Marketing: The Case Against One AI Lab
By Matt Cullerton, founder of Mavric and Apso
Adam Aldred, who runs a marketing agency on his own, argues that marketers are betting too heavily on Anthropic and ignoring OpenAI's Codex. In a 63-second TikTok (@adam_aldred) he says Codex computer use and a workflow called Threadmaxing save him more time than anything Anthropic offers in the browser. The features he names are real and shipped, but the video shows no demo, so his time savings are a claim to test yourself.
Why does marketing AI content skew toward Anthropic?
Aldred opens by asking whether he is the only marketing person using Codex, because the posts he sees online only talk about Anthropic. He asks whether that content dominates because it gets engagement or because people get better results with it. He does not answer the question.
His point is that the one-sidedness is itself a warning sign. When a whole discipline converges on one vendor's tools, the convergence can reflect social proof rather than tested performance.
Should you commit to one frontier lab?
His core advice is plain: being married to only one frontier lab is a mistake. He qualifies it in both directions. Do not spend all your time chasing every new release, and do not overcommit to a single vendor either. Set aside a fixed share of your time for testing alternatives.
For a solo operator who bills client hours, that testing time has a real cost. That is why his report that it paid off carries some weight. Engineers apply the same logic to cloud vendors, and he applies it to AI tools for SEO and paid ads work.
What is Codex computer use?
Aldred says computer use in Codex is better than anything Anthropic is doing in the browser. OpenAI shipped this in the Codex app for macOS and Windows. The app includes an in-app browser the agent can click, type into and inspect.
OpenAI also released a Chrome extension that lets Codex work inside the user's logged-in browser sessions, with tools like Gmail and Salesforce. For agency work, that means the agent can operate in the same accounts a marketer already uses.
What is Threadmaxing?
Threadmaxing is a term for attaching recurring wake-up schedules to a Codex thread. The agent returns to the same conversation on a timer, checks for changes and prepares work for a human to approve.
Aldred concedes the pattern could probably be replicated on Anthropic's tools, but says it is a lot easier on Codex. He credits computer use and Threadmaxing together for letting him manage a few dozen clients across SEO and paid ads with no staff.
What the video does not show
The whole video is a single talking-head shot. There is no screen recording, no dashboard and no demonstration of Codex or Threadmaxing. Every claim about time saved and feature quality is stated rather than shown.
Aldred offers follow-up posts on how he runs the workflow for his clients. Until those exist, the video is a pointer to two Codex features worth trying, and his comparison with Anthropic is his own judgment.
Bottom line
Codex computer use, the Chrome extension and scheduled thread automation are real, shipped features, and the sources below describe them. Aldred's claim that they beat Anthropic's browser tools for marketing work is unverified. His broader advice, to keep a slice of time for testing a second vendor, costs little to follow.
Sources
- Codex by OpenAI - product page for OpenAI's coding and computer-use agent.
- Introducing the Codex app - OpenAI's announcement of the Codex desktop app.
- Codex for (almost) everything - OpenAI's post on computer use, the in-app browser and plugins.
- OpenAI Codex arrives in the browser with new Chrome extension - The New Stack on Codex working inside logged-in Chrome sessions.
- Codex-Maxxing shows how AI can support long-running workflows - explainer on the thread automation pattern behind Threadmaxing.
Published August 21, 2026. Based on a TikTok by @adam_aldred. Full video notes (markdown)