Vibe coding gets your app built. Monitoring helps you keep it working.

Croft has added built-in monitoring and telemetry to every app in its workspace, requiring no SDKs or code changes. The company says logs, errors, and telemetry are automatically surfaced to a user's AI so it can diagnose failed requests and slow performance with runtime evidence rather than code alone.

AI has made it much easier to go from an idea to a working app. You describe what you want, try the result, and keep iterating until it does what you had in mind. But once other people start using it, you face a different question: If you’re new to building software, this is a good time to get familiar with monitoring and telemetry. They give you visibility into what your app does after you share the link—and give your AI useful evidence when something needs fixing. Testing your app yourself matters. Sign up, submit the form, save some data, and check the main flows. But your testing won’t cover everything. Someone might submit an empty field, open a link to a record that no longer exists, or trigger a request that takes much longer than it did during your testing. Sometimes they’ll report a problem. Often, they’ll just leave. Monitoring helps you spot errors and slow requests so you can investigate without relying entirely on user reports. These terms can sound more complicated than they are: The value comes from connecting them. Knowing a request failed is useful. Seeing what happened around that failure can help explain why. You don’t need to watch a dashboard all day. You need enough visibility to notice problems and enough context to investigate them. When you build with AI, your first response to a bug will probably be to ask the AI to fix it. That makes sense. But there’s a difference between reading the code and seeing what happened when the code ran. Imagine a user tells you that saving their profile failed. Several things could explain it: unexpected input, a database problem, or a request to another service that didn’t complete. If all you tell the AI is “saving profiles is broken,” it has to work out which possibility to investigate. An error from the failed request and the logs around it give the investigation a much stronger starting point. The same applies to performance. “My app is slow” is a broad problem. “This request is taking several seconds” is something you can investigate. A useful workflow is: That last step matters. A plausible code change still needs verification. You can also make your prompts more specific: “Investigate this error. Explain what the logs show, identify the likely cause, and fix it.” “Look into this slow request. Identify where the time is being spent and suggest an improvement.” Good telemetry helps the AI answer those questions with evidence from your app. We’ve added built-in monitoring and telemetry to every app in your Croft workspace. There are no SDKs to install and no code changes to make. Monitoring lets you see errors and slow requests and investigate their causes. Logs, errors, and telemetry are automatically made available to your AI, giving it the context and tools to help diagnose problems and work on fixes. For someone just getting started, that removes a setup step that’s easy to postpone. You can build your app and ask your AI to investigate its behaviour in the same workspace. You can read more on the Croft monitoring page https://oncroft.net/monitoring/ . Getting your first app into people’s hands is a milestone. Monitoring helps you take the next step: understanding their experience and keeping the app working as you improve it.