'%3e%3crect%20x='128'%20y='122'%20width='263'%20height='88'%20rx='36'%20fill='%232878ef'/%3e%3c/g%3e%3cg%20transform='rotate(7%20237%20258)'%3e%3crect%20x='128'%20y='215'%20width='216'%20height='82'%20rx='34'%20fill='%23f58220'/%3e%3c/g%3e%3cg%20transform='rotate(-5%20259%20350)'%3e%3crect%20x='128'%20y='304'%20width='264'%20height='88'%20rx='36'%20fill='%23ed438b'/%3e%3c/g%3e%3c/svg%3e){kind=small}
Which AI agents work with Enjoy?
Enjoy works with five coding agents from different AI companies. See what each one needs, how they differ, and how to pick one to start with.
An agent is a program from an AI company that reads your project, edits its files, and runs commands to build what you ask for. Enjoy installs agents on your computer, runs them for you, and gives them all the same message box, so you don’t need to learn each one’s commands.
The five agents
| Agent | Made by | You sign in with |
|---|---|---|
| Claude Code | Anthropic | Your Claude account |
| Codex | OpenAI | Your ChatGPT account |
| Grok Build | SpaceXAI | Your Grok or X account |
| OpenCode | Anomaly | No single account: it uses the AI providers you set up in it |
| Antigravity | Your Google account |
How they differ
- Claude Code has six effort levels, from Low to Ultracode, and a Fast speed on supported models. It has six permission modes, from Plan to Bypass permissions, and it can use your Chrome browser.
- Codex has six effort levels, from Low to Ultra, and a Fast speed. Its four permission modes range from Read-only to Full access, and its usage can include credits.
- Grok Build has four effort levels, from Low to Extra high, and three permission modes: Plan, Auto, and Always approve. It has no Fast speed. See Use Grok Build.
- OpenCode uses models from the providers you set up in it, including free and local models. Each model uses its own effort setting, and it has two modes, Build and Plan. Enjoy can’t show its usage. See Use OpenCode with free, local, or your own models.
- Antigravity has four effort levels and two permission modes. Enjoy can’t show Antigravity’s approval requests, so by default it runs without asking. See Use Antigravity.
Claude Code, Codex, and Grok Build can keep several accounts signed in at once, and each project can use a different one. See Switch between agent accounts.
What every agent shares
- You talk to every agent in the same message box. Their replies, questions, and approval requests look the same.
- Every agent gets Enjoy’s own tools, so it can ask you questions, write docs, save workflows, and show you a preview of what it built. See What agents can do inside Enjoy.
- Each reply shows the name of the agent that wrote it. Point to a reply to see the model and effort it used.
- Agents run on your computer, inside your project’s folder.
Choosing one to start with
- Start with an agent you already have a plan for. Some agents use a subscription, others work with models you’ve configured, and Enjoy won’t buy or change a subscription for you. See Do I need a separate AI subscription?
- You can install several agents and choose one for each conversation, or switch agents partway through. When one plan runs low, another agent can carry on.
- New conversations start with the agent and settings you used last in that project.
On the web app and iPhone
Agents always run on your computer, even when you send a message from somewhere else. Installing agents, signing in, managing accounts, and choosing permission modes happen in the Enjoy app on your computer.
- In the web app, you can switch between the agents installed on your computer and choose a model, effort, and speed. You can’t switch to an agent that would get broader permissions than the conversation already has.
- The iPhone app uses the agent chosen on your computer. You can choose the model, effort, and speed for your next message.