Floyd is an enterprise-level world model that actually learns how you'll use a computer and learns exactly which steps you'll take to do certain tasks, so it does them as you would.
You can train it to mimic you so that every action taken would be like you are doing it.
I'm trying to crack the human-like use of computers with AI
Report
🔌 Plugged in
Hey @tirell_arzu super glad I stumbled upon this! Next step in automation for sure. Love the performance dashboard showing confidence score and analytics. Wishing you all the best with the launch 🚀
Questions:
How do you determine next actions to take for the model in terms of predicting what comes next?
Is there a fallback option on the off chance the model incorrectly defined and carried out steps but you would want to reverse a task?
The next action is predicated based on training data captured from people using a computer. The model then goes into the simulation and keeps learning, so for right now, it's doing its own trial and error until it gets it right
Currently, there is no fallback option, but thanks for that i'll place that guardrail there for that.
Report
💡 Bright idea
Looks cool man! Is it mainly for enterprises or for smaller teams as well?
The concept is genuinely interesting for enterprise automation but one question your procurement reviews will surface early:
What level of system permissions does Floyd need to operate? And is the training data the "how you do tasks" layer isolated per org, or is there a shared model that learns from all users' workflows?
For enterprise buyers, those two questions come before the demo. Getting clear answers on them now would also help you close faster when security teams get involved.
It operates in isolated, self-hosted environments, so nothing leaves the company's infrastructure. The deeper the persmission the better the results ofcourse as it will be able to learn more.
So it won't be a shared model, but one isolated to the organization.
To train the base model I am using the data from my job simulation platform for entry-level talent at https://anterpise.com/
From there, it's all isolated environments to learn within a specific organization
Report
@tirell_arzu That isolation architecture is the right foundation for enterprise one shared model learning across orgs would be a dealbreaker for most procurement teams before you even got to the demo. The self-hosted angle is a strong differentiator too. One follow-up: how are you handling the permissions scope when Floyd is initially onboarding into a new org? That first access grant tends to be the moment security teams push back hardest curious how you've thought through making it a smooth handoff.
Replies
Floyd enterprise world model
Hey @tirell_arzu super glad I stumbled upon this! Next step in automation for sure. Love the performance dashboard showing confidence score and analytics. Wishing you all the best with the launch 🚀
Questions:
How do you determine next actions to take for the model in terms of predicting what comes next?
Is there a fallback option on the off chance the model incorrectly defined and carried out steps but you would want to reverse a task?
Floyd enterprise world model
@minhajulll Hey man thanks for the response
The next action is predicated based on training data captured from people using a computer. The model then goes into the simulation and keeps learning, so for right now, it's doing its own trial and error until it gets it right
Currently, there is no fallback option, but thanks for that i'll place that guardrail there for that.
Looks cool man! Is it mainly for enterprises or for smaller teams as well?
Floyd enterprise world model
@abhinavramesh Hey man, it can be used for both. I can give you a 3-day guest pass to try it out unrestricted
@tirell_arzu Sure, can you send to abhinav@matterhorn.so
The concept is genuinely interesting for enterprise automation but one question your procurement reviews will surface early:
What level of system permissions does Floyd need to operate? And is the training data the "how you do tasks" layer isolated per org, or is there a shared model that learns from all users' workflows?
For enterprise buyers, those two questions come before the demo. Getting clear answers on them now would also help you close faster when security teams get involved.
Floyd enterprise world model
@avinash_matrixgard Hey man, thanks for the reply
It operates in isolated, self-hosted environments, so nothing leaves the company's infrastructure. The deeper the persmission the better the results ofcourse as it will be able to learn more.
So it won't be a shared model, but one isolated to the organization.
To train the base model I am using the data from my job simulation platform for entry-level talent at https://anterpise.com/
From there, it's all isolated environments to learn within a specific organization
@tirell_arzu That isolation architecture is the right foundation for enterprise one shared model learning across orgs would be a dealbreaker for most procurement teams before you even got to the demo. The self-hosted angle is a strong differentiator too. One follow-up: how are you handling the permissions scope when Floyd is initially onboarding into a new org? That first access grant tends to be the moment security teams push back hardest curious how you've thought through making it a smooth handoff.