Kavela vs Claude's cloud agents

1. Your own model plan

How Claude does it

Every Claude cloud agent runs on Claude. Anthropic's docs put it plainly for parallel work: in every approach the workers are Claude sessions ⁷. Cloud sessions always use your own subscription credentials and are not available through third-party providers such as Bedrock or Vertex ^{1, 8}. Pro costs $20 a month and Max starts at $100 a month ¹⁵.

Anthropic also sets the terms for other platforms. Its Claude Code legal page does not allow third parties to route requests through a person's Free, Pro or Max plan on their behalf, but it does not stop a person signing in to the unmodified Claude Code program with their own subscription, including where a platform hosts it ¹⁴.

How Kavela does it

Kavela runs on your choice of plan, agent by agent. With your own Claude plan, the unmodified Claude Code runs inside the agent's own cloud computer and you sign in to it yourself, which is the form Anthropic's terms describe. With a ChatGPT plan, the same agent can run Codex instead, and Kavela's included Kav AI usage covers anyone who has neither.

Because the choice is per agent, one team can mix Claude and Codex agents, and you keep paying for the subscriptions you already have.

You keep the Claude plan you chose, and the work it does stops being limited to one repository and one person.

2. Managing a team of agents

How Claude does it

Projects are Anthropic's answer to running many agents at once. The coordinator takes what you send, decides what becomes a thread and follows each thread to the end; the Overview groups threads by state and has tabs for files, pull requests and routines ⁵. The limit is 200 new threads a day across your projects, and a thread limit you ask for is treated as a preference ⁵.

Projects are in public beta on the Pro and Max plans only, not yet on Team or Enterprise, and the docs warn that a Pro plan reaches its usage limit sooner on days you run one ⁵. Routines belong to your individual account and are not shared with teammates ⁴. The docs describe no calendar of agents' work; the Overview is a list of threads ⁵.

How Kavela does it

In Kavela the lead agent is Kav. You tell Kav what you want, in the workspace chat or in Telegram, and Kav plans it, hands each part to the right worker agent, follows up and brings back the result. The workers are a standing team: named agents with their own skills, connected apps and run history, and you can hire a new one by describing the job.

Every task, from every agent, sits on one calendar with its start time, due time and live state, so a week of agent work reads like a week of a team's work: what is scheduled, what is running, what needs you and what is done.

A list of threads tells you what is happening now; a calendar tells you what your agents did last week and what they will do tomorrow.

3. Agents that can sign up for things

How Claude does it

The Claude Code docs describe no email address of the agent's own ^{1, 7}. Cloud sessions also cannot sign in to a website interactively: the environment docs list browser-based logins such as AWS SSO as not supported, because they need a browser that cannot run in a cloud session ³.

How Kavela does it

Every Kavela workspace has its own receive-only inbox, Kavela Mail. When an agent signs up for a service on your behalf, it uses a fresh address from that inbox, waits for the verification email, reads the code or the link and carries on. The wait shows in Approvals, and received mail is deleted after 30 days.

With a live browser sign-in for the services that need one, agents can create the accounts a job requires and set those services up, rather than stopping to ask you to do it.

Most real setup work starts with an account and a verification email; an agent that can get past that does far more of the job.

4. What agents need, and what they deliver

How Claude does it

In a project, the Overview's Waiting on you group lists threads that need your reply or approval. The approval itself stays inside its thread: the docs note that telling Claude in the project conversation to go ahead does not reach it, and the thread waits until you answer it there ⁵. Files the threads produce go to a Library tab and pull requests to a Pull requests tab ⁵.

Claude's general cloud sessions ask before each action by default, with Manual, Auto and Skip modes, and send the question to your phone when a decision is needed ^{10, 11, 12}.

How Kavela does it

Kavela collects everything your agents need into one Approvals inbox for the whole workspace. Each request is a typed card you answer where it is: connect an app, sign in to a site in a live browser, approve a post before it goes out, choose between options, or let an agent know the email it was waiting for has arrived. Other work carries on while a card waits.

What agents produce arrives as deliverables attached to the task that made them, from documents and reports to sites, listed under Delivered and opened in one click.

When every request is a card in one place, checking on ten agents takes as long as checking on one.

5. Many agents, running on schedule

How Claude does it

Each Claude Code cloud session runs in an isolated virtual machine with up to 4 CPUs, 16 GB of memory and 30 GB of disk ³. Routines run on a schedule from hourly to weekly, with a minimum interval of one hour, and an account can run 100 scheduled runs an hour ⁴. Routines are a research preview, and one turns itself off after 72 hours without a GitHub connection ⁴. A green run status only means the session started and stopped without an infrastructure error, not that the task succeeded ⁴.

How Kavela does it

Each Kavela agent works in its own cloud computer, with a shell, files, git and a browser, which starts when work starts and sleeps when it stops. One agent's work never slows another's, and work keeps going after you close the tab. The workspace's files persist between runs, so tomorrow's run starts from today's output.

Automations run on a schedule or on an event, such as a webhook or a merged pull request, and can be limited to an active window with start and end dates. There is no cap on how many are switched on at once.

Recurring work is only useful if it is shared, visible and judged by its result, not by whether a machine started.

6. Connecting your tools

How Claude does it

Claude Code cloud sessions need GitHub to clone a repository and open a pull request; a repository from GitLab or Bitbucket can be uploaded, but the session cannot push results back ^{1, 2}. Sessions and routines can use the MCP connectors you added in claude.ai, which on Team and Enterprise plans only admins can add ⁹. Network secrets for other APIs are available on Pro and Max, not yet on Team or Enterprise ³.

How Kavela does it

Kavela connections belong to the workspace: connect a service once and every agent can use it. A service can be connected by OAuth sign-in, as an MCP server, with an API key, with custom secrets, or by signing in to it in a live browser you control. Passwords and keys go into the card, never through the model.

You can also leave it to the agent. With Let Kav set it up, the agent reads the service's documentation, works out the right way to connect it and does the setup, creating the account first if there is none.

The tools a business runs on rarely all live on GitHub; an agent team has to reach the rest of them.