👋 Hey there, tech enthusiasts!
I'm Sarvar, a Cloud Architect with a passion for transforming complex technological challenges into elegant soluti...
For further actions, you may consider blocking this person and/or reporting abuse
After a week testing their preview, I am putting this ghost back in the containment unit. While I see the vision, here are the friction points I encountered:
1) Rigid workflow kills momentum Kiro enforces a waterfall approach: Requirements → Design → Task List → Coding. While this structure works for building features from scratch, it becomes a bottleneck during iteration.
The real issue? Requirements and design artifacts don't auto-update when you make changes through prompting. Your implementation and specs drift apart unless you manually update each step - a time sink that breaks the iterative flow essential to productive coding sessions.
2) Poor codebase comprehension Kiro struggled to analyze my existing code and conventions. It missed reusable components that should have been leveraged with minor tweaks. Its rigid task-list execution leaves no room for the coding agent to adapt once implementation begins.
In contrast, Claude Code dynamically adjusts its approach as it discovers new implementation details during execution.
I'm sticking with Claude Code's planning mode for now, and it's unlimited Max plan. Until Kiro can keep specs synchronized with ongoing changes and actual code, it adds overhead rather than enhancing productivity.
Thanks for taking the time to share such detailed feedback it's incredibly valuable. I'm part of the AWS Community Builders program and have direct access to the Kiro development team, so I’ll make sure to pass this along to them. Your insights, especially around iterative workflow challenges and spec-code drift, are much appreciated and will help shape future improvements. Thanks again for contributing to the conversation!
Great summary, really interesting to see how this works!
Unfortunately, I’m still stuck on the waitlist so haven’t been able to try Kiro myself yet. In my own custom AI agent setup, I’ve found the agent control loop, specs-driven development, and especially a clear Task List to be essential to keeping the agent on track and autonomous.
If you’re curious, I wrote a post about my own findings how I got a single agent to ship production-ready code with no mid-task tweaks. Here’s Part 1 of the series.
Would love to hear your thoughts if you get a chance to check it out!
Thanks for sharing your experience that sounds fascinating! I totally agree that the agent control loop and task list clarity are key for any kind of autonomous progress. I’d love to dive into your Part 1 and see how you approached specs-driven development and kept your agent on track end-to-end. It’s always inspiring to see real-world applications outside the mainstream tools like Kiro. Once I check it out, I’ll definitely share my thoughts. Appreciate you dropping this here!
Any update on wish list? Didn’t receive any mail regarding access to the Kiro?
Aws team is working on the access part you will receive mail soon. Stay Tune!
Thank You!
La app tiene una lista de espera qui sierra saber cuánto tiempo di lata para que te notifique para poder descargarla app
Yo esperé más o menos una semana en la lista de espera antes de recibir el código. Una vez que lo recibes ya puedes descargar y usar la app sin problema.
Great comparison! I’ve been testing Cursor lately and I have to admit it feels way more intuitive than I expected.
True! But Kiro have more GenAI and Agentic AI capabilities 😉
En kiro quiero saber cuado sale la opción de paga.
Todavía no han anunciado una fecha exacta para la versión de paga. Por ahora Kiro está en fase de acceso anticipado con invitación, y los planes de pago se habilitarán en próximas actualizaciones.