This product is pretty crazy. I gave it some backe...
# random
s
This product is pretty crazy. I gave it some backend code and it's deploying a web app for me. (we both get bonus credits if you use the referral link) https://manus.im/invitation/FT95GD3S9WRW1P9
šŸ’Ŗ 1
t
Just trying to understand. I work in tech but I am not a traditional dev. You gave it pure backend code and it wrote the front-end for you?
How will you validate that the front end does what you expect it to?
And did it just write the front end or is it actually deploying it to some infra/cloud?
My background is more automation and Platform Engineering. Right now, I’m mostly dealing with Ansible, Terraform and K8s on Azure. And I cannot get AI to produce anything usable, other than helping me with Terraform loops or blocks I need quickly generated.
Hope it’s OK that I’m asking.
s
I devloped the code in a jupyter notebook, and Manus adapted it to Flask. It adapted my code turn it into a working backend and also wrote frontend. It did some testing, which produced a caught exception that it ignored. I told it to fix the problem, which it did, and then retested and confirmed the problem was fixed. Now it's working on deploying to some kind of public endpoint, but I don't know the details of how yet.
I will be doing extensive testing on my own once deployed to see that the app functions like it should, and I'm sure there will be some bugs, but it did like 90% of the work
t
When you say to some public endpoint - does it just pick one at random or did you provide your cloud account?
And when you say it fixed the problem, did you verify it only fixed that problem and didn’t also make unrelated changes?
^ I’ve had issues with the above
s
I didn't have to provide any account info for the deployment. It works quickly and autonomously, so it's hard to monitor every change it makes in real time. I'm letting it do it's thing and then I will assess afterwards
I believe it is deploying on its own server or cloud environment and then sharing a public link
I haven't experimented with using it within my own environment yet.
t
Now I’m curious about how many user apps it’s running in its own environment, how it’s isolating them, how much it costs, etc.
c
Wait until you ask it to change the color of a button and it deletes 3 fields from a form before you celebrate šŸ˜‰
šŸ˜‚ 1
t
I feel silly that everyone is saying how awesome it is and how it does everything end to end, but I have a lot of problems replicating that claim.
However, I am not a developer, I’m on the platform side. My job is to turn pre-sales ā€œhallucinationsā€ into reality…
s
it's not 100% there and definitely still needs troubleshooting. But it still did a lot of the work and saved a lot of time
r
ā€œNever trust anything that can think for itself if you can’t see where it keeps its brain?ā€œā€ -> In this case the brain happens to be in China
My point is that the security of these systems is pretty rough
t
I manage Kubernetes clusters for a financial firm. I’d love to deploy some of the cool looking AI tools to help the developers whose apps run in the cluster get help/answers/etc, but I DO NOT TRUST these tools in terms of what AI it’s using and where the data is being sent.
And when I say this, people act like I’m the crazy one.
I do need to spend time researching options, deploying the model that’s being queried in clusters, etc. But the point is I’m genuinely concerned about how comfortable people are deploying random junk they find online into live and sensitive client environments.
s
I'm using this for a hobby project, so no high security needs
g
What's the difference between this and all the other AI development tools? bolt, lovable, v0, etc?
r
They all have issues but this one is China-based