Ask HN: Are cloud coding agents useful in real workflows yet? | Hacker News Reader

Ask HN: Are cloud coding agents useful in real workflows yet?

7 pointsRperry21743 comments

Objectively, cloud-based agents have several functional advantages over local ones: persistent environments, asynchronous execution, longer-running tasks, shared context across teams, easier access to compute, and the ability to operate while you’re offline or doing something else.

Yet a lot of discussion around AI coding tools still seems centered on editor UX, autocomplete, and whether something is “just” an IDE, GUI, or TUI wrapper.

Especially as many major tools are now investing in cloud agents it makes me wonder if people are underestimating where the real leverage may be or if cloud just isn't actually "the future".

Are people here actually using cloud coding agents in real workflows yet?

View on HN

I haven't seen any use cases around me, although I'm not that advanced user.

But there are a few things that makes me understand why it's not used more:

- It's way too expensive for me to use it in private life/side project

- Data protection and integration into companies is lacking for company use

- They are very, very badly advertised. I still have to watch several YT videos and hope for reddit comments, to figure out what they mean as an "agent", what it can do, and does it really only work with github.

I am recently catching up on a lot of agent-related topics and I still struggle to understand what they are selling.

I think cloud coding agents are useful, but they still face an adoption problem: most users optimize for immediacy, not maximum capability.

If a workflow needs account setup, billing, permissions, remote environments, or context handoff, many people drop before seeing the value.

Local tools usually win on trust, latency, and “open it and start now.” Cloud tools win when tasks are long-running, collaborative, or need scalable compute.

My guess is both models stay relevant, but the tools that remove the most friction will win.

I think the issue currently is, that you can easily give your local agent instance access to a lot of tools you just have installed on your machine. A cloud agent is a blank canvas (which can be a good thing imo). This makes setting them up more work. -> People want something that works out of the box. Another issue is probably that cloud agents have a datacenter IP and get blocked a lot.

Reply on news.ycombinator.com