Documentation overview

Chrome extension

The Chrome extension puts The AI CMO inside your browser: a sidebar agent on any page, one-tap research and competitor tools, and – most importantly – it lets chat work through your browser with your existing logins instead of a cloud browser.

Install

The extension installs from the app (not the Chrome Web Store yet): download the ZIP from the extension page, unzip it, open chrome://extensions, switch on Developer mode, choose Load unpacked, and select the folder. Pin "The AI CMO" via the puzzle-piece icon. About two minutes, and the in-app guide walks every step. Updates work the same way – download the new ZIP and reload.

Signing in: the extension uses your existing session – just keep a signed-in theaicmo.com tab open. If the sidebar says "Sign in Required", open theaicmo.com and sign in; that's the whole fix. The site editor can also sign in on its own, so it keeps working with no app tab open – see Connecting this browser, under Site editor below.

The sidebar

Open the side panel on any page. Two tabs:

  • Agent – a chat with your AI CMO that can see the page you're on and act on it. Attach files or pull from your library. Each message is priced on what it uses, like chat.
  • Features – one-tap tools for the page you're viewing: Save to Research (clip the page as a source), Watch this competitor (track the site for changes), Preflight selected text, an AI Brand Checker (how AI models perceive any brand), Ad Style Transfer (generate ads from any image on any website), and your image gallery.

Right-clicking anywhere adds the same powers to the context menu: draft in an editable field, rewrite a selection in your brand voice, tighten copy, preflight before publishing, save images to your gallery, clip research, or add a site to competitor watch.

Site editor

The site editor is where you build site tests and site changes – on your own website, by pointing at what you want to change. It needs the Website SDK on the site.

From the sidebar. Open your site and the side panel. The Site editor card shows who you are signed in as, the site that matches the page (its own address, a site the page is under, such as shop.example.com on example.com, or the site whose SDK a staging address runs; a picker when several match), whether the Website SDK was found on the page, and the site's tests with their status. Edit opens a test in the editor on this page; Open in app shows it on Experiments. New test asks for the type – A/B, Multivariate or Site change – and a name, makes a draft and opens it.

From the app. On Experiments, Edit on site opens the test's start page in a tab of this browser with the editor on it. When the extension is missing or too old for the editor, the app says so and links to the download.

In the editor. It docks on the right of your page. Pick a version tab, then click any element in Edit mode to change its text, swap an image, hide it, restyle it, change a link, or make clicks on it a goal; Browse mode lets you use the page normally to reach menus, pop-ups and signed-in pages. Changes save to the test's draft as you work, and nothing reaches a visitor until you publish. The editor follows you as you move between pages of the site, and closes when you close it or the tab. Experiments covers versions, goals, preview links, publishing and results in full.

While the editor is open on a page, the Website SDK leaves your tests off there, so you always edit the page as it is.

Connecting this browser

Connect in the sidebar (or /extension/connect in the app) gives the extension its own sign-in in this browser, as you – press Connect this browser once. The site editor then works with no app tab open and reaches only the brands and sites you can reach in the app. Connect only a browser profile that is yours: anyone using it can use the extension as you.

Browser automation

Give the agent a task in plain words – "log my leads from this page into the CRM", "fill in this form from my brand profile" – and it plans and executes the steps: navigating, clicking, typing, extracting data, working across tabs. While it works, the window shows a glowing border and a banner ("Browsing in progress – do not interact with this window"); let it finish or cancel from chat.

It's built to be honest about limits: it stops at paywalls and dead ends rather than guessing, retries at most a few times, and it will ask you to handle logins, captchas, and payments itself – it never does those for you.

Why it changes chat

With the extension installed, chat on theaicmo.com shows Extension connected – and browser tasks from chat run in your browser, with your existing logins to Gmail, HubSpot, LinkedIn, your ad platforms, your CMS. Without it, the same tasks fall back to the built-in cloud browser, which asks for consent, costs 40 credits per session (refunded if the session fails), and needs a login the first time it meets each platform. The extension route is both more capable and far cheaper.

Permissions, plainly

The extension asks for broad permissions because automation needs them: access to the sites it works on, tabs, and Chrome's debugger – which is how it types and clicks like a person. Two things you'll notice and shouldn't worry about:

  • During automation Chrome shows its standard bar: "The AI CMO started debugging this browser." That's the input mechanism, and it goes away when the task ends.
  • Because the extension is developer-mode installed, Chrome may occasionally ask you to confirm you want to keep it.

Your saved credentials never leave the encrypted vault, and the agent only acts when you give it a task.