Kavela vs Gemini's cloud agents

1. Your own model plan

How Gemini does it

Spark runs on Gemini 3.5, and Google's help pages describe no other model or provider ^{1, 2}. It needs a Google AI Pro or Ultra subscription; Ultra costs $100 or $200 a month ^{2, 12}. Jules also runs on Gemini models, with a free tier and higher limits included in Google AI Pro and Ultra ^{10, 11}.

The Gemini agent for Gemini Enterprise goes further on models: Google says it works across the Gemini family and Anthropic's Claude models today, with other models to come ³. The announcement describes no way to bring a model subscription you already pay for ³.

How Kavela does it

Kavela runs on your choice of plan, agent by agent. With your own Claude plan, Claude Code runs inside the agent's own cloud computer and you sign in to it yourself; with a ChatGPT plan, the same agent can run Codex instead. The sign-in is yours: Kavela does not route requests through anyone's plan itself, and its included Kav AI usage covers anyone who has neither plan.

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

Google's enterprise agent is announced with more than one model; Kavela lets each agent run on a model plan you already pay for.

2. Managing a team of agents

How Gemini does it

Spark is one agent. You give it tasks, schedules and skills, which Google calls the what, when and how, and follow them in a Tasks list and a Schedules page grouped into Ongoing, Paused and Completed ^{2, 7}. At launch Google said Spark would be able to create custom sub-agents; the official release log, whose latest entry is 17 July 2026, lists no such release ^{1, 5}.

The Gemini agent announced for Gemini Enterprise is built around many agents. It can create a roster of sub-agents for a job, each with its own identity, running parallel and sequential steps for hours or days, and a coworker agent with a standing role can delegate and coordinate tasks; colleagues work with it by adding it to a Chat space or mentioning it ³. Google has not stated when this will be available ³.

Jules runs several coding tasks at once, but you open each one to check on it and approve its plan ¹³. The Spark and Jules docs describe no calendar of agent work; schedules and tasks are lists ^{7, 13}.

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, by day, week or month, 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.

Google has announced standing coworker agents without a date; in Kavela the team exists now, and its week of work reads on one calendar.

3. Agents that can sign up for things

How Gemini does it

Google has announced agent email for enterprises. In the Gemini agent announcement, coworker agents get their own Workspace account, with an email address on the company's agents domain, a calendar, Drive and a place in the company directory ³. That identity lives in your company's Google Workspace, and Google has not stated when it will be available ³.

For Spark, Google said at launch that people would be able to email Spark directly; the official release log does not list it yet ^{1, 5, 14}. Spark works with your own Gmail, where it can search, draft, reply and apply labels ². When a website asks Spark's remote browser to sign in, the task stops until you take over and sign in yourself; with Chrome auto browse, Spark can use your logged-in accounts and saved passwords instead, in the US first and with your device and Chrome running ^{2, 8}.

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. Kavela Mail only receives: agents use it to sign up, and do not send email from it.

Google's coworker agents are announced with Workspace accounts of their own; Kavela's inbox does a narrower job, receiving sign-up emails, and it is live now.

4. What agents need, and what they deliver

How Gemini does it

Spark asks for your review and confirmation before it sends communications, changes your data, makes purchases or submits web forms, and you take control yourself for passwords or payment details ². Planned edits to shared Docs, Sheets and Slides wait for your review, and since July 2026 its notifications include more detail when input is needed ^{2, 5}. What it produces is Google Workspace files you can edit in Canvas ². The help pages describe each confirmation inside its own task, and no inbox that gathers them across tasks ².

Gemini Enterprise has the closest thing to a shared approvals inbox. Its Inbox is a central place to monitor, guide and manage all agent activity, with notifications sorted into groups such as Needs your input, Errors and Completed, and its Agent Designer supports human-in-the-loop checkpoints ⁶. Google announced these in April 2026 and said they would roll out to customers over the following months ⁶.

Jules asks you to approve its plan before it changes any code, and delivers a branch or a pull request ^{4, 9}.

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 see that an agent is waiting for an email. 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.

Gemini Enterprise sorts agent requests into groups; Kavela makes each one a card you answer where it sits: connect, sign in, approve or choose.

5. Many agents, running on schedule

How Gemini does it

Spark works in a remote browser and on a remote computer that runs code ². Its schedules run at set times, from once to yearly, or watch Gmail or a topic and act when something matches ⁷. You can have up to 15 tasks running at once, and a schedule does not run while 15 are already running; you can keep up to 50 active schedules ^{2, 7}. Google notes that monitors are not ideal for fast-moving data or time-critical tasks ⁷.

Gemini Enterprise describes long-running agents that work for up to days at a time in Google's cloud sandboxes, and the new Gemini agent runs inside an Agent Sandbox with its own network boundary; you can schedule its work or have it respond to events ^{3, 6}. Jules runs each task in its own virtual machine, with 3, 15 or 60 tasks at once depending on the plan, and supports scheduled tasks ^{10, 13, 15}.

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 each run starts where the last one finished, for every agent on the team.

6. Connecting your tools

How Gemini does it

Spark works across Google Workspace (Calendar, Docs, Drive, Gmail, Keep, Sheets, Slides and Tasks), Google's search services and YouTube, plus several third-party apps and custom apps ². Since June 2026 you can connect a personal or third-party app by its MCP server address ^{5, 16}. For websites it relies on the remote-browser hand-over and, in the US, on Chrome with your saved logins ^{2, 8}. The help pages describe no API keys or custom secrets ².

The Gemini agent announcement says it can work with any MCP server inside or outside your company network and carries the agent's identity through standards such as OAuth; Gemini Enterprise lets admins connect custom tools through BYO-MCP ^{3, 6}. Jules works with the GitHub repositories you allow, environment variables and six curated MCP servers added by API key ^{11, 13}.

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.

Spark is at its best inside Google's apps; a business also runs on tools that need a key, a secret or a sign-in, and an agent team has to reach those too.