BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//pretalx//cfp.nluug.nl//nluug-najaarsconferentie-2025//speaker//SW
 8AEV
BEGIN:VTIMEZONE
TZID:CET
BEGIN:STANDARD
DTSTART:20001029T040000
RRULE:FREQ=YEARLY;BYDAY=-1SU;BYMONTH=10
TZNAME:CET
TZOFFSETFROM:+0200
TZOFFSETTO:+0100
END:STANDARD
BEGIN:DAYLIGHT
DTSTART:20000326T030000
RRULE:FREQ=YEARLY;BYDAY=-1SU;BYMONTH=3
TZNAME:CEST
TZOFFSETFROM:+0100
TZOFFSETTO:+0200
END:DAYLIGHT
END:VTIMEZONE
BEGIN:VEVENT
UID:pretalx-nluug-najaarsconferentie-2025-GB8KQP@cfp.nluug.nl
DTSTART;TZID=CET:20251120T105000
DTEND;TZID=CET:20251120T113500
DESCRIPTION:Can large language models move beyond theory and conversation i
 nto the domain of autonomous penetration testing? In this talk\, we presen
 t the results of a series of cutting-edge experiments aimed at evaluating 
 and enhancing the real-world offensive capabilities of LLMs through agenti
 c architectures.\n\nWe explore whether automating reconnaissance improves 
 LLM performance\, how model alloys inspired by XBOW research can outperfor
 m individual SOTA models\, and why in many cases—less is more. Our work 
 integrates practical tools like Burp Suite and custom MCP pipelines\, test
 ing multi-agent groupchat architectures with specialized roles—planner\,
  recon\, exploit—and ultimately evolving into swarm-based methodologies 
 that unlock collaborative intelligence.\n\nDespite advances\, we found tha
 t most high-performing tool calls still default to simple Python or curl\,
  raising key questions about true integration maturity. Our architecture d
 emonstrates how a well-orchestrated LLM-agent system can rival traditional
  pentesting frameworks in specific tasks\, pass certain benchmarks\, and u
 ncover real-world vulnerabilities.\n\nThis talk shares technical insights\
 , lessons learned\, and the future potential for AI-augmented security tes
 ting.
DTSTAMP:20261103T082205Z
LOCATION:Room 1
SUMMARY:Beyond the Prompt: Architecting Agentic LLMs for Autonomous Securit
 y Testing - Joost Grunwald
URL:https://cfp.nluug.nl/nluug-najaarsconferentie-2025/talk/GB8KQP/
END:VEVENT
END:VCALENDAR