Grok Bot vs Claude Code: Set Up Each For Its Job (2026)

Grok bot vs claude code is a choice between a general agent that lives on a cloud computer and a coding agent that lives inside your repo, so I set the two up in completely different orders.

Grok Bot gives you named Bots that work through a browser, a terminal and connected apps, per the Grok Bot overview.

Claude Code is Anthropic's agentic coding tool: it reads a codebase, edits files and runs commands, according to the Claude Code overview.

Below is the exact order I'd switch things on for each one, the settings I'd change first, and the gotchas both sets of docs warn about.

Neither xAI nor Anthropic publishes a comparison of the two, so every side-by-side call on this page is my own reading of each vendor's docs, checked on 5 November 2026.

Want my team to map this onto your business? Book a free SEO strategy session and we'll walk through it with you.

What each tool is actually built for

Grok Bot is built for business work that happens in websites and apps: research, inbox chores, spreadsheets, reports and anything behind a login.

You message a Bot, give it the job and the access it asks for, and it comes back when something needs your OK.

Claude Code is built for software work: building features, fixing bugs and automating development tasks across a whole codebase.

It can touch the web too, through its Chrome integration, but the repo is its home.

My one-line version: Grok Bot is an operator for business chores, and Claude Code is an engineer for code.

If you're weighing Grok Bot against OpenAI's coding agent instead, my Grok Bot vs Codex setup guide covers that split.

The operator's comparison table

Setting	Grok Bot	Claude Code
Built for	General business work across apps and websites	Coding in a repository
Where work runs	One cloud computer shared by all your Bots, each Bot on its own screen	Your own machine by default; cloud sessions run in an isolated VM each
Your own computer	Off-limits unless you allow execution (Ask every time by default)	The default workspace
Approvals	Auto Review, approval cards and Ask first rules	Six permission modes plus allow, ask and deny rules
Recurring work	Routines on a schedule or on events	Cloud scheduled tasks, Desktop scheduled tasks, /loop, GitHub Actions
Memory	Per Bot, plus a few shared facts about you	CLAUDE.md you write, plus auto memory Claude writes
Extensions	Skills, apps, MCP servers	Skills, subagents, hooks, MCP servers, plugins
Model	No picker; Grok Bot chooses	You choose with /model
Platforms	macOS, Windows, Linux, iOS, Android	Terminal, VS Code, JetBrains, desktop, web, mobile

Step 1: Confirm which plan unlocks each tool

For Grok Bot, xAI's X Subscription page says Premium, Super and Ultra include it, and Plus doesn't.

The overview adds that legacy paid Cursor plans and Cursor Teams include it, and that you can link a SuperGrok or X Premium+ subscription.

For Claude Code, the advanced setup page lists Pro, Max, Ultra, Team, Enterprise or an Anthropic Console account.

The free claude.ai plan does not include Claude Code.

You can also point Claude Code at Amazon Bedrock, Google Cloud's Agent Platform or Microsoft Foundry instead of a Claude plan.

Step 2: Install, then decide where the work runs

Grok Bot comes as a desktop and phone app, and the Bots do their work on a cloud computer rather than your laptop.

The computer and apps page is blunt about the model: every Bot on your account uses the same computer, with shared cookies, files and command-line credentials.

Each Bot gets its own screen on that shared computer, and one Bot runs one computer-use task on its screen at a time.

Gotcha: the screens are work surfaces, not security boundaries, so never park a credential there that another Bot shouldn't use.

Claude Code installs with one command on macOS, Linux or WSL:

curl -fsSL https://claude.ai/install.sh | bash

Windows users run the PowerShell installer from the same quickstart, or use the desktop app, which includes Claude Code.

Out of the box it works on your machine, in whatever project folder you start it from.

Cloud sessions are the exception: the cloud sessions page says each one runs in an isolated, Anthropic-managed virtual machine and clones your repo from GitHub.

Start one from claude.ai/code, the Claude app, the desktop app or claude --cloud in the terminal.

🔥 Want both agents set up without the guesswork? Inside the AI Profit Boardroom you get step-by-step video trainings, the installable Agent OS, four coaching calls a week and 3,400+ members comparing setups like this one. → Join the AI Profit Boardroom

Step 3: Set approvals before the first real task

Grok Bot's approvals page describes Auto Review, which checks each action before the Bot runs a command, uses an app or touches its computer.

Most actions run on their own, and the ones that need you show an approval card with Allow once, Always allow and Deny.

Anything matching an Ask first rule always gets a card.

I write boundaries into the first message too: no sending, no publishing, no purchases, no deleting and no permission changes without asking.

For your own laptop, go to Settings, Computer, Your computer and check Execution on this computer, which defaults to Ask every time.

The docs suggest Never allow unless a Bot has a specific reason to touch local files.

Claude Code uses permission modes, listed on the permission modes page, and Shift+Tab switches between them in the CLI.

Manual (config value default) asks before most edits and commands, acceptEdits lets file edits through, and plan blocks edits until you approve a plan.

auto has a classifier review actions in the background, dontAsk denies anything not pre-approved, and bypassPermissions is for isolated containers and VMs only.

From v2.1.283, auto mode is the built-in starting mode for interactive terminal and VS Code sessions, so check which mode you're in before you start.

Deny rules block in every mode, which makes them the safest place to fence off anything you never want touched.

On an unfamiliar repo I start in plan mode, read the plan, then move to Manual.

Step 4: Give each tool its memory

Claude Code's memory page describes two layers: CLAUDE.md files you write, and auto memory notes Claude writes from your corrections.

Both load at the start of every session, and /memory opens them.

Put build commands, coding standards and the things you'd otherwise re-explain into CLAUDE.md.

The docs add a key caveat: memory is context, not enforced configuration, so use a PreToolUse hook when something must be blocked every time.

Grok Bot memory lives with each Bot, according to the Bots page, while a few facts such as your name and time zone are shared across all your Bots.

There's no memory screen for a personal Bot, so ask it directly: "What do you remember about me?"

Logins, cards and contact details go in Vault, not memory.

Step 5: Add skills, MCP servers and hooks

In Claude Code, a skill is a SKILL.md file you call with /skill-name, per the skills docs.

Subagents run side jobs in their own context window, and hooks run your shell commands at fixed points, such as formatting code after every edit.

MCP servers connect it to trackers, databases and APIs; the MCP guide covers /mcp.

In Grok Bot, you save a skill from a task that worked, or use Teach a task to demonstrate a browser workflow for up to ten minutes, per the skills and routines page.

Apps come from Connect apps in the sidebar, and a Bot can add an MCP server by URL or by a command that runs on its computer.

Grok Bot documents no user-level hooks; the closest feature is inference hooks, an Enterprise beta that sends model requests to your organization's security server for an allow or deny verdict.

Step 6: Automate last, with routines or scheduled tasks

Grok Bot routines run on a schedule written in plain words, or on a trigger: a Slack message, a GitHub, Linear, Sentry or PagerDuty event, an email to the Bot's address, or a webhook.

They run in the cloud while your laptop is closed.

A new routine waits for its next slot, so press Test to run it now, and remember a test does real work.

Email a routine writes waits as a draft until you press Send.

Claude Code gives you three schedulers, compared on its scheduling page:

Cloud scheduled tasks (formerly routines, in research preview): a schedule, an API POST with a bearer token, or GitHub events; minimum interval one hour; a fresh clone, so no local files. Details on the scheduled tasks page.
Desktop scheduled tasks: run on your machine with your files, but only while the app is open and the computer is awake.
/loop: repeats a prompt inside an open session, for quick polling.

Then there's Claude Code GitHub Actions: run /install-github-app, then mention @claude in a pull request or issue.

Naming gotcha: both products have used the word routines, and Anthropic's admin toggle can still read Routines even though the feature is now scheduled tasks.

Step 7: Reach both from your phone

The Grok Bot mobile app on iPhone and Android connects to the same Bots, routines and shared cloud computer as the desktop app.

Use it to approve cards and check results away from your desk.

Claude Code reaches your phone three ways, per the platforms page: cloud sessions in the Claude app's Code tab, Remote Control for a session running on your machine, and Dispatch to the desktop app on Pro, Max and Ultra.

On those same three plans, mentioning @Claude in Slack can turn a bug report into a pull request.

Where Grok Bot wins

Work that lives behind website logins, because sign-ins persist on its computer.
Event triggers outside code: email, Slack, Linear, Sentry, PagerDuty and webhooks.
Several named Bots working in parallel and handing off in group chats.
Non-technical teammates, since setup is a message, not a config file.

Where Claude Code wins

Anything in a codebase: features, fixes, refactors and tests.
Fine-grained control, from six permission modes to deny rules and hooks.
Running where your code already is: terminal, VS Code, JetBrains or the desktop app.
CI work through GitHub Actions and GitLab CI/CD, and headless runs with claude -p.
Model choice, from Haiku for light jobs to Fable for the long ones.

How I'd run the two together

Give each tool the half of the job it was built for, and let GitHub be the handoff point.

Claude Code makes the code change in your site repo and opens a pull request.
A Grok Bot routine watches that single pull request, a trigger the Grok Bot changelog added in v0.56.0.
The Bot reads the PR comments through its GitHub connector and posts a plain-English summary in your chat.
After you merge, the same Bot opens the live page in its browser and lists anything that looks broken.
Anything it finds goes back to Claude Code as the next task.

📞 Want a second pair of eyes on your agent setup? Book a free SEO strategy session and my team will look at which of your jobs belong with each tool. → Get your free strategy session

Verdict by job type

Pick Grok Bot if the job is inbox triage, research, spreadsheets, reports or chores in web apps, and nobody on the team writes code.

Pick Claude Code if the job ends in a commit: a website theme, a script, an integration or a test suite.

Run both if you have a small codebase and a pile of business admin around it.

We share the exact prompts we use for splits like this inside the AI Profit Boardroom.

Three small-business setups, step by step

A hair salon on Grok Bot

Give one Bot its own email address from the Email app and send booking enquiries there.
Create a routine that runs when email arrives at that address.
The Bot reads the enquiry, checks your booking site in its browser and drafts a reply with open slots.
The draft waits until you press Send, because routine email always waits.

An online shop on Claude Code

Open the shop's theme repo in the terminal and run claude.
Start in plan mode and ask for a new size-guide section on product pages.
Approve the plan, review each edit in Manual mode, and let Claude run the theme's tests.
Add a CLAUDE.md line with the build command so the next session starts faster.

A small agency on both

Share one Grok Bot with the team and connect Slack.
Add a Slack trigger for the phrase "site fix" in the client channel.
The Bot collects the request, screenshots the page and writes a clear brief.
A developer pastes the brief into Claude Code, which makes the fix and opens a pull request for review.

Also On Our Network

FAQ: Grok Bot vs Claude Code

Can Grok Bot write code like Claude Code?

Grok Bot's cloud computer has a terminal and a GitHub connector, so it can run commands and read repos. Claude Code is the tool built for editing a codebase, with permission modes, hooks and CI integrations made for that job.

Does Claude Code run on my computer or in the cloud?

On your computer by default, inside the project folder you start it from. Cloud sessions, started from claude.ai/code, the Claude app or claude --cloud, run in an isolated Anthropic-managed VM each.

Which permission mode should I start Claude Code in?

Use plan mode on a repo you don't know yet, then Manual to review each action. Keep bypassPermissions for isolated containers and VMs, as the docs say.

Do my Grok Bots each get their own computer?

No. All your Bots share one cloud computer with shared files, logins and credentials, and each Bot gets its own screen on it.

Are Grok Bot routines the same as Claude Code routines?

No. Claude Code renamed its routines to cloud scheduled tasks, which run on a schedule, an API call or GitHub events. Grok Bot routines run on schedules or on Slack, GitHub, Linear, Sentry, PagerDuty, email and webhook triggers.

📺 Video notes + links to the tools 👉

🎥 Learn how I make these videos 👉

🆓 Get a FREE AI Course + Community + 1,000 AI Agents 👉

About Julian

I'm Julian Goldie, an SEO entrepreneur, author and founder of the AI Profit Boardroom, which has 3,400+ members.

I help business owners scale with AI agents, automation and SEO.

I've built a 7-figure SEO and link building agency, Goldie Agency, with a team of 70+ SEO experts.
I've grown a YouTube channel to 400,000+ subscribers.
I wrote "Link Building Mastery".
My Udemy courses have taught over 29,000 students.

→ Get my best AI training inside the AI Profit Boardroom

Install one, give it one narrow job, and book a free strategy session if you want my team to settle grok bot vs claude code for your business.