Y
HN Search
Hacker News Search
new
|
comments
|
top
|
jobs
mahmoud-almadi
searching PlanetScale…
1.
▲
2.
▲
3.
▲
4.
▲
5.
▲
6.
▲
5 ms
·
1.
▲
by
mahmoud-almadi
1y ago
Thanks for scheduling a demo! Looking forward to chatting soon. You can definitely monitor this remotely by looking at the logs of the runs which includes screenshots, but we don't currently support live streaming of the desktop screen
2.
▲
by
mahmoud-almadi
1y ago
You make valid points. At the end of the day we're focused on delivering real value while taking every security precaution we can reasonably take and build new technology at the same time. Eventually as we grow we'll be able to do
3.
▲
by
mahmoud-almadi
1y ago
I appreciate your skepticism. At the end of the day we're focused on delivering real value while taking every security precaution we can reasonably take and build new technology at the same time. Eventually as we grow we'll be abl
4.
▲
by
mahmoud-almadi
1y ago
Great point! The issue with hotkeys is a series of hardcoded keyboard shortcuts is likely to fail due to unexpected popups which are quite frequent. Cyberdesk can automate entire workflows with keyboard shortcuts only, and we typically push
5.
▲
by
mahmoud-almadi
1y ago
Not yet! Vision only has been doing pretty well so far, but we're definitely looking into that at some point. Thanks for the suggestion!
6.
▲
by
mahmoud-almadi
1y ago
You are correct in that ByteDance did releas UI-TARS which sounds like a really good open source computer use model according to some articles I read. You could run that locally. We haven't tested it so I wouldn't know how it perf
7.
▲
by
mahmoud-almadi
1y ago
I don't know too much about training your own computer use model other than it would probably be a very hefty, very expensive task. However, I believe ByteDance released UI-TARS which is an excellent open source computer use model acco
8.
▲
by
mahmoud-almadi
1y ago
These points you're making are excellent! Self hosting is inevitably a part of our roadmap. Cyberdesk will have a future where we host our entire agentic framework on your own servers. AI models and the whole backend included. I can to
9.
▲
by
mahmoud-almadi
1y ago
Thanks! Really appreciate the awesome thoughts. 1) You're totally right about this problem! We handle this issue with intelligent caching and heavy prompt/context engineering. These measures have been controlling agent behavior pr
10.
▲
by
mahmoud-almadi
1y ago
Thanks for the breakdown Seth! We did indeed get their SOC 2 Type II reports and made sure they showed no significant security vulnerabilities that will impact our usage of their service.
11.
▲
by
mahmoud-almadi
1y ago
Right now we are taking the policies we signed with our LLM vendors as a verification of a zero data retention policy. We did also get their SOC 2 Type II reports and they showed no significant security vulnerabilities that will impact our
12.
▲
by
mahmoud-almadi
1y ago
Great point! Yes you are correct in that the actual "agent" lives in the cloud and its actions are executed by a proxy running on the desktop. Hopefully at some point we can set up a straightforward installation procedure to have
13.
▲
by
mahmoud-almadi
1y ago
sorry it came off that way! could you elaborate on that thought?
14.
▲
by
mahmoud-almadi
1y ago
We haven't looked into this kind of approach yet, but definitely worthwhile to do at some point!
15.
▲
by
mahmoud-almadi
1y ago
Good point. In a way we can verify to a customer that we have that policy set up with them by showing them the certificate. But you are correct in that we haven't gone as far as asking for proof from Anthropic or OpenAI on not retainin
16.
▲
by
mahmoud-almadi
1y ago
Are you referring to the LLM being used or where the actions (click, type, etc) are being executed? The actual actions can be executed on any windows machine, so the actual execution can take place locally on your device. The LLMs we'r
17.
▲
by
mahmoud-almadi
1y ago
We sign BAAs with all our healthcare customers + all our vendors. Currently using Claude computer-use. Zero-data retention signed with both Anthropic and OpenAI, so none of the information getting sent to their LLMs ever get retained
18.
▲
Launch HN: Cyberdesk (YC S25) – Automate Windows legacy desktop apps
73 points
by
mahmoud-almadi
1y ago
|
63 comments