Post
The new security research model can retrieve and analyze relevant information more efficiently, synthesizing insights to provide greater context
Introducing Searchlight AI Agent 2.0, our most capable model yet for cybersecurity investigations. Using the new model, Searchlighters can retrieve and analyze relevant information more efficiently and synthesize insights to gain greater context to support their investigations. It also supports document memory within a chat session, allowing it to reference previous points shared in the same conversation.
Key model improvements
- New dark web search query helper tool, providing starter questions to support your investigations
- Boosted information retrieval speed through optimized agent-to-tool separation
- Increased agent accuracy with new query context tools for more targeted results.
- Improved answer fidelity by assigning one agent per searchable index
- Maintain conversational context across multi-turn investigations with the addition of chat memory, eliminating the need to repeat prior steps or information
Early success stories
A US State Agency was alerted to a threat actor who claimed to have breached its unemployment-benefit accounts. Using the Cerberus AI Research Assistant, the analyst was able to query details from the initial post and instantly locate the post’s original dark web source. The AI assistant tool translated the post, generated a synopsis of the thread, and extracted the key details, for the wider criminal investigation team.
By pivoting on these AI-derived insights, the analyst was able to secure pivotal, new details about the threat actor’s identity and operational methodology. This verified intelligence was delivered to the state’s cybersecurity office, significantly accelerating the criminal investigation and enabling rapid, pre-emptive defense measures to protect public data.