Crowdbotics uses AI to analyze your code base across three layers of abstraction, giving developers, architects, and business stakeholders the insights they need to modernize and maintain legacy code.
FWIW the site will be undergoing major changes soon which should help more clearly explain it.
Well, that's my point. I see some JSON screenshots with highlighting, but nothing really that corresponds with what my own brown- and grayfield experience leads me to look for, and not really anything very persuasive toward the interpretability of what is on offer, so that I can evaluate the quality of the tool's "judgment." But I suppose that just goes to clarify the case for the redesign you mentioned!
Not that you're asking, but what would really convince me would be being able to point a cut-down version of the tool, presumably in a BYOAPI fashion, at one of my own old stale repos like https://github.com/aaron-em/same-encoder, and see what kind of a modernization plan it comes up with and how that compares with my own. A toy problem to be sure, but what better field for evaluation? Think of it like a take-home interview, perhaps.
Feel free to email me, dan at crowdbotics dot com. I'm on vacation this week but could run your repo next week. Right now we're targeting older languages like COBOL and Java but it does work well with newer languages like JS since there's a lot of it in the training data.
Oh, okay, that makes a lot more sense, in an application where engineers are scarce already and getting more so, and where the kind of "deep time" it takes for a safe, well-understood refactor of a complex running system probably also isn't much on offer. Good grief, I couldn't seriously imagine taking on a refactor of a mature COBOL application especially, what with how those tend to grow in place over decades - it's ambitious as anything and I respect it! Enjoy your vacation.
Comments
There is a description at the top of the site:
FWIW the site will be undergoing major changes soon which should help more clearly explain it.
Well, that's my point. I see some JSON screenshots with highlighting, but nothing really that corresponds with what my own brown- and grayfield experience leads me to look for, and not really anything very persuasive toward the interpretability of what is on offer, so that I can evaluate the quality of the tool's "judgment." But I suppose that just goes to clarify the case for the redesign you mentioned!
Not that you're asking, but what would really convince me would be being able to point a cut-down version of the tool, presumably in a BYOAPI fashion, at one of my own old stale repos like https://github.com/aaron-em/same-encoder, and see what kind of a modernization plan it comes up with and how that compares with my own. A toy problem to be sure, but what better field for evaluation? Think of it like a take-home interview, perhaps.
Feel free to email me, dan at crowdbotics dot com. I'm on vacation this week but could run your repo next week. Right now we're targeting older languages like COBOL and Java but it does work well with newer languages like JS since there's a lot of it in the training data.
Oh, okay, that makes a lot more sense, in an application where engineers are scarce already and getting more so, and where the kind of "deep time" it takes for a safe, well-understood refactor of a complex running system probably also isn't much on offer. Good grief, I couldn't seriously imagine taking on a refactor of a mature COBOL application especially, what with how those tend to grow in place over decades - it's ambitious as anything and I respect it! Enjoy your vacation.