Hey Product Hunt 👋
We built Devin because we wanted an AI engineer that actually fits how developers work, not another tab to babysit. Today we're launching Devin for Terminal. Devin lives right in your CLI, alongside the tools you already use. And when a task is bigger than a single session, you can hand it off to a cloud agent and keep moving.
Same Devin. Two surfaces. Pick whichever matches the moment.
Would love your honest feedback. What would make this a daily driver for you?
the keep-the-task-alive-after-i-close-my-laptop pattern is the right unlock. the gnarly part is making the cloud session's environment match enough of mine that 'it worked locally' isn't a coin flip when it comes back. how are you handling state parity between the cli and the cloud agent?
Report
Benchmarks look impressive… real test is how it handles edge cases and messy requirements.
Report
devin is the one i keep coming back to think about because the scope of what it's doing is just different from other coding tools. it's not autocomplete, it's not "generate this function" — it's taking an issue and working through the whole repo to ship a PR. the part that makes me curious is how it handles ambiguity in requirements, like when the ticket is underspecified and a human dev would normally ask three clarifying questions before touching anything. does it make assumptions and push forward or does it pause and check?
Report
As someone learning DS, the annoying part is always setting up environments and debugging random dependency errors before even starting. If Devin in terminal can just handle that grunt work autonomously I can actually focus on the analysis. Curious how it does with Python/pandas setups specifically.
Replies
Mind the Graph
OpenPlugin
the keep-the-task-alive-after-i-close-my-laptop pattern is the right unlock. the gnarly part is making the cloud session's environment match enough of mine that 'it worked locally' isn't a coin flip when it comes back. how are you handling state parity between the cli and the cloud agent?
Benchmarks look impressive… real test is how it handles edge cases and messy requirements.
devin is the one i keep coming back to think about because the scope of what it's doing is just different from other coding tools. it's not autocomplete, it's not "generate this function" — it's taking an issue and working through the whole repo to ship a PR. the part that makes me curious is how it handles ambiguity in requirements, like when the ticket is underspecified and a human dev would normally ask three clarifying questions before touching anything. does it make assumptions and push forward or does it pause and check?
As someone learning DS, the annoying part is always setting up environments and debugging random dependency errors before even starting. If Devin in terminal can just handle that grunt work autonomously I can actually focus on the analysis. Curious how it does with Python/pandas setups specifically.