
What I like best about Kestra is that it changes the way I think about backend automation and system design. It is not just a tool for running tasks; it gives a complete orchestration layer where I can design, execute, monitor, debug, and improve workflows in a structured way.
Before using Kestra, most automation ideas felt like they needed a custom backend server, cron jobs, queues, retry logic, logging, secret handling, state management, and separate monitoring. With Kestra, a lot of that infrastructure becomes part of the workflow itself. I can define the behavior declaratively, connect APIs and services, add triggers, manage secrets, handle retries, inspect outputs, and monitor executions from one place.
The best part for me is the execution visibility. When a workflow runs, I am not blindly checking terminal logs or guessing where something failed. Kestra shows the execution step by step through the UI, logs, task outputs, topology view, Gantt view, and execution state. If something fails, it is much easier to identify the failed task, fix the issue, and continue debugging with context.
I also like how useful Kestra is for AI and automation-heavy systems. I used it while building projects like AutoPR Engine, DevAlert, and Sentinel Grid. These projects involved GitHub webhooks, API calls, AI model integrations, parallel execution, Telegram and Gmail notifications, scheduled workflows, local persistence, and live workflow monitoring. Kestra made it easier to connect all these moving parts without turning the project into messy backend glue code.
Another thing I appreciate is that Kestra works well for both learning and real implementation. I started with Kestra Academy and local Docker setup, but quickly moved into real workflows with multiple flows, namespaces, triggers, retries, and integrations. That made the learning practical instead of only theoretical.
Overall, the best thing about Kestra is that it helps move from scattered scripts to reliable, observable, production-style workflows. It lets me focus more on system behavior and less on rebuilding the same backend orchestration logic again and again. Review collected by and hosted on G2.com.
What I dislike about Kestra is not really a single feature, but the initial learning curve when you are coming from normal scripting or backend development.
At the beginning, there are many concepts to understand together: flows, tasks, triggers, inputs, outputs, namespaces, secrets, plugins, retries, executions, logs, scheduling, and the execution lifecycle. If someone is new to workflow orchestration, it can feel a little overwhelming because Kestra is not just asking you to write code — it asks you to think in terms of workflows and system behavior.
Another challenge is that for small automations, Kestra can feel more structured than necessary at first. If the task is very simple, writing a quick script may feel faster in the beginning. But once the workflow grows and needs retries, monitoring, scheduling, multiple integrations, or failure handling, Kestra starts to make much more sense.
I also feel that examples and beginner-friendly real-world templates could be even more helpful, especially for AI workflows, webhook-based automations, notification pipelines, and multi-flow projects. The documentation is useful, but when building more complex systems with multiple namespaces, APIs, secrets, and parallel tasks, having more production-style examples would make onboarding faster.
Overall, these are not deal-breakers. The learning curve is expected because Kestra solves a serious orchestration problem. But for beginners, the first few workflows require patience and hands-on practice before the real value becomes obvious. Review collected by and hosted on G2.com.
