Looking forward to sharing the stage at [un]prompted with the wizard himself, @hrbrmstr, as we showcase "Orbie" (a custom-built AI agent that analyzes internet-scale honeypot data to surface emerging threats and even identify campaigns).

We’ll share what works, what doesn’t, and the specific campaigns we caught that traditional methods missed. You’ll see how domain expertise embedded in tooling enables LLMs to operate on billions of network sessions, and why that matters more than the model you choose.

https://unpromptedcon.org/

Agenda - [un]prompted

[un]prompted
@ntkramer @hrbrmstr Very cool! Will you be open sourcing Orbie or is it proprietary?

@keith_lawson @ntkramer We're going to do a series of posts about the impetus, genesis, and architecture.

Orbie is pretty tied to our internal stack, but I'm/we're def going to provide enough info for defenders to replicate w/their stacks.

i've open sourced a few of the MCPs already too, but they'll be documented better when we blather more about it.

@hrbrmstr @ntkramer Excellent! I'm currently working on RL agents that conduct SOC tasks and would be interested to see if I can combine that with your work.