Inspiration

Cybersecurity teams face overwhelming challenges: chaotic meetings, misleading brainstorming sessions, and significant delays in threat understanding—up to 287 minutes just to get started. With increasing cyberattack costs ($700M in 2017 alone), there’s an urgent need to support experts more intelligently and efficiently.


What it does

Sherlock is an AI-powered cybersecurity assistant that helps experts investigate incidents faster by combining up-to-date knowledge (like CVEs and news), internal documentation, and expert insights. It acts as a tireless co-pilot, guiding investigations and helping with remediation—without replacing human experts.


How we built it

We developed Sherlock by integrating large-scale internal documentation, real-time cybersecurity updates, and expert reasoning into a single smart assistant interface. The system is optimized to reduce the time needed for incident comprehension and response.


Challenges we ran into

  • Making sense of vast, unstructured internal documentation
  • Keeping up with rapidly evolving cybersecurity news and vulnerabilities
  • Designing a system that assists rather than overwhelms or replaces human experts

Accomplishments that we're proud of

  • Reduced average investigation time by 73%
  • Successfully demonstrated how AI can augment, not replace, cybersecurity professionals
  • Built a demo that resonates with real-world incident response needs

What we learned

  • Cybersecurity workflows benefit greatly from context-aware AI assistance
  • Experts need fast, trustworthy tools—not just more data
  • A clear, intuitive interface is critical for adoption in high-stakes environments

What's next for Sherlock

  • Expand real-world testing and integrations with SOC tools
  • Improve natural language understanding for complex incident queries
  • Add collaborative features to assist teams, not just individuals

Built With

Share this project:

Updates