Can't let this go by without a comment. You are directionally right. The challenge continues to be the fragmented IT infrastructure at enterprises and nation states. All vulnerabilites that aren't patched, all configurations that havent been updated to ensure that doors are open, can and will be found with Mythos class models. (Ad: we just announced that we can bring Mythos testing to you exclusively or as part of a multi harness model). There needs to be a concerted effort by all enterprises and nation states to discover and have a remediation plan. Having said that, the story will not stop there. Attack paths will be developed by AI models trained to exploit vulnerabilities, to launch phishing attacks, to hijack credentials. However robust your perimeter defense is and patching strategy is, one needs an AI based detection and remediation approach in case attackers get past. The solve here is a robust enterprise security data lake, running machine speed detection and remediation. Don't fall for band aid approaches, of which there are plenty, fragmented data layers with agents running on top (broadly called AI SOC agents) won't do the trick. AI needs high fidelity data to operate, it needs proactive detection and agentic remediation across all enterprise assets. Not to add more - while we are busy doing this, the deployment of agents and agentic AI is rapidly creating new attack surfaces because most coding and agentic deployments today aren't secured.