Uploaded March 2026 | Updated September 2026, 3 weeks ago
I caught up with Rishi from ProjectDiscovery to check out Neo, their new AI platform that unleashes autonomous security agents to map your attack surface and actually validate vulns for you. We talked about how this approach is completely crushing false positives , and you've definitely gotta see the live demo he showed me!
What we talk about:
The Evolution of ProjectDiscovery:
Moving from traditional open-source cybersecurity tools to building an entirely AI-native platform called Neo.
Mapping the Attack Surface:
How Neo uses deployed security agents to continuously map out an organization's internal and external infrastructure to identify exploitable vulnerabilities.
Crushing False Positives:
Why traditional tools struggle with noise and how Neo uses a completely separate validation step in a dedicated sandbox to reduce false positives by over 90%.
Managing AI Limitations:
The technical hurdles of building LLM-based security tools, like preventing context window overload and teaching the AI to use search tools dynamically instead of stuffing its working memory.
Inside the Platform:
A look at how users can hand off goals to autonomous agents that automatically plan their own subtasks, run on a daily schedule, and push verified reports straight to ticketing systems like Jira or Linear.
00:00 - Catching up: What have you been building at ProjectDiscovery?
03:18 - What primary problems are customers using Neo to solve?
06:59 - How is Neo tackling the classic issue of false positives?
14:22 - How do you manage AI context window limits over long periods?
17:16 - Can customers provide specific environmental or threat modeling context?
19:04 - Beyond vulnerability backlogs, what other use cases is Neo handling?
21:43 - How does the AI approach chain smaller subtasks and tools together?
28:53 - Live Demo: Can we see the interface and how a task runs?
37:04 - Can multiple agents report to a shared location for a unified review?
39:37 - How can people sign up, and how does the pricing model work?
Subscribe to the newsletter at:
danielmiessler.com/subscribe
Join the UL community at:
danielmiessler.com/upgrade
Follow on X:
https://x.com/danielmiessler
Follow on LinkedIn:
linkedin.com/in/danielmiessler
I caught up with Rishi from ProjectDiscovery to check out Neo, their new AI platform that unleashes autonomous security agents to map your attack surface and actually validate vulns for you. We talked about how this approach is completely crushing false positives , and you've definitely gotta see the live demo he showed me!
What we talk about:
The Evolution of ProjectDiscovery:
Moving from traditional open-source cybersecurity tools to building an entirely AI-native platform called Neo.
Mapping the Attack Surface:
How Neo uses deployed security agents to continuously map out an organization's internal and external infrastructure to identify exploitable vulnerabilities.
Crushing False Positives:
Why traditional tools struggle with noise and how Neo uses a completely separate validation step in a dedicated sandbox to reduce false positives by over 90%.
Managing AI Limitations:
The technical hurdles of building LLM-based security tools, like preventing context window overload and teaching the AI to use search tools dynamically instead of stuffing its working memory.
Inside the Platform:
A look at how users can hand off goals to autonomous agents that automatically plan their own subtasks, run on a daily schedule, and push verified reports straight to ticketing systems like Jira or Linear.
00:00 - Catching up: What have you been building at ProjectDiscovery?
03:18 - What primary problems are customers using Neo to solve?
06:59 - How is Neo tackling the classic issue of false positives?
14:22 - How do you manage AI context window limits over long periods?
17:16 - Can customers provide specific environmental or threat modeling context?
19:04 - Beyond vulnerability backlogs, what other use cases is Neo handling?
21:43 - How does the AI approach chain smaller subtasks and tools together?
28:53 - Live Demo: Can we see the interface and how a task runs?
37:04 - Can multiple agents report to a shared location for a unified review?
39:37 - How can people sign up, and how does the pricing model work?
Subscribe to the newsletter at:
danielmiessler.com/subscribe
Join the UL community at:
danielmiessler.com/upgrade
Follow on X:
https://x.com/danielmiessler
Follow on LinkedIn:
linkedin.com/in/danielmiessler
