The workflow decision
For screenshot-driven browser control, the useful comparison is whether Notte or Kernel can meet the following outcome: The screenshot contains the required controls and the selected action produces the expected website state, with a saved image explaining the choice.
Rendering fidelity and screenshot latency are part of the workload. CDP compatibility alone does not establish that another browser engine produces equivalent screenshots or coordinates.
A proposed acceptance case
An automation chooses its next action from a screenshot. Text extraction alone cannot establish whether the page rendered the controls it needs.
This is an evaluation design, not a claim that this workflow has been executed successfully on every provider.
| Part of the test | Required observation |
|---|---|
| Controlled input | A controlled page containing a visible button, a layout-dependent target and an expected final visual state. |
| Pass condition | The screenshot contains the required controls and the selected action produces the expected website state, with a saved image explaining the choice. |
| Negative case | Present a page where the target is obscured or not rendered. The workflow must stop or request help rather than claim that a text-only result proves success. |
| Evidence to retain | Input frame, selected target, post-action frame and an independent website-state check. |
Documented mechanisms, side by side
These labels describe product mechanisms. The linked source and the acceptance case determine whether a mechanism meets this workflow's requirement.
| Requirement | Notte | Kernel |
|---|---|---|
| Browser control | Documented | Documented |
| Saved browser state | Documented | Documented |
| Live view | Documented | Documented |
| Deployment choice | Enterprise options | Check deployment offering |
Sources: docs.notte.cc [1] ↗docs.notte.cc [2] ↗docs.notte.cc [3] ↗docs.notte.cc [4] ↗kernel.sh [5] ↗kernel.sh [6] ↗kernel.sh [7] ↗kernel.sh [8] ↗
Inspecting a run
Notte: Live view shows an active browser and can be embedded in an application. Recordings let you review a finished session. Function runs have their own history and results, so keep the run ID alongside the browser session ID.
Kernel: Live browser URLs support an embedded view and a read-only option. Replays and browser telemetry are separate observability surfaces; app invocation logs add execution context.
Sources: docs.notte.cc [1] ↗kernel.sh [2] ↗
Browser control
Notte: Create an owned browser session, connect Playwright or Puppeteer over CDP, and stop the session when the work finishes. Browser lifecycle management remains necessary even when the control socket disconnects.
Kernel: Kernel exposes browser creation and control APIs. Its documentation distinguishes on-demand browsers from configured browser pools, so compare the acquisition path your application will actually use.
Sources: docs.notte.cc [1] ↗kernel.sh [2] ↗
Deployment ownership
Notte: Use a remote browser from your existing process or deploy the browser automation as a Notte Function. These choices put lifecycle, scheduling and monitoring responsibilities in different places.
Kernel: Kernel Apps move the automation process next to the browser. Alternatively, an external process can control a browser through the browser API. Choose based on execution locality and deployment ownership.
Sources: docs.notte.cc [1] ↗kernel.sh [2] ↗
Pricing model
Notte: The public usage rates are $0.01 per browser-hour, $0.05 per Function-hour and $10 per GB of managed proxy traffic. Paid plans return their price as usage credits. Price the relevant meters and plan eligibility for the workload.
Kernel: Use the plan and browser configuration in Kernel's pricing documentation. Managed networking, account allowances and app execution make a browser-hour comparison only one part of the decision.
Sources: notte.cc [1] ↗kernel.sh [2] ↗
What should change the choice?
Evaluate Notte's browser and inspection path on the actual visual workflow. A documented CDP connection and a fast lifecycle are useful starting points, but neither is evidence of screenshot parity.
Kernel: Kernel's browser-local Apps make execution placement part of the decision. Compare the repeated action path with the proposed Notte controller or Function deployment before concluding that a lifecycle or hourly-rate difference decides the workflow.
Prefer the current engine when the visual workflow depends on behavior already validated there. A faster non-equivalent engine can invalidate the action-selection method.
Sources: docs.notte.cc [1] ↗docs.notte.cc [2] ↗docs.notte.cc [3] ↗notte.cc [4] ↗kernel.sh [5] ↗kernel.sh [6] ↗kernel.sh [7] ↗kernel.sh [8] ↗
The cost boundary
Use the same attempted work and eligible account capacity in both estimates. The browser-hour rate does not include every proxy, execution or identity meter, and the cost of an incomplete job is still part of the workflow budget.
The workload evidence should include: Input frame, selected target, post-action frame and an independent website-state check.
Sources: notte.cc [1] ↗kernel.sh [2] ↗
Make a bounded pilot
Use the same viewport and pages, verify the intended state after each action, and include screenshot transfer time in task latency.
Use the same authorized account, input and final-state check in each configuration. Treat a missing product contract as an open question; do not turn an undocumented capability into an unsupported claim or a successful test result.
- Save baseline screenshots and expected website state from a controlled account.
- Change the browser provider while keeping the visual controller constant.
- Review mismatched frames and actions before allowing account-changing operations.