公开(公告)号：US20240427631A1

公开(公告)日：2024-12-26

申请号：US18475058

申请日：2023-09-26

Applicant: CrowdStrike, Inc.

Inventor： Dragos Corlatescu ,  Dan Belibov ,  Daniel Radu ,  Theo Chihaia ,  Marian Radu

IPC: G06F9/48 ,  G06F9/54 ,  G06N3/045

Abstract: Systems and methods for incremental solves using LLMs for API calls is presented. The systems and methods produce, by a first large learning model (LLM), a processing plan based on a first prompt, wherein the processing plan includes a plurality of tasks corresponding to a plurality of services. The systems and methods send a plurality of messages corresponding to the plurality of tasks to a plurality of service agents, wherein the plurality of service agents correspond to the plurality of services and comprise a plurality of second LLMs that produce a plurality of agent responses. The systems and methods then generate a query response based on the plurality of agent responses.

公开(公告)号：US20240427807A1

公开(公告)日：2024-12-26

申请号：US18461305

申请日：2023-09-05

Applicant: CrowdStrike, Inc.

Inventor： Dragos Corlatescu ,  Dan Belibov ,  Daniel Radu ,  Elena Budau ,  Marian Radu

IPC: G06F16/332 ,  G06F16/335 ,  G06F40/284

Abstract: The present disclosure produces a first output in response to inputting a first prompt into a large language model (LLM). The first prompt comprises a first document group that corresponds to a second document group, and the LLM is limited by a maximum token limit that is less than a token count of the second document group. The present disclosure generates a second prompt that comprises a subset of the second document group corresponding to the first output. The present disclosure then produces a second output based on the subset of the second document group in response to inputting the second prompt into the LLM.

公开(公告)号：US20250139233A1

公开(公告)日：2025-05-01

申请号：US18494509

申请日：2023-10-25

Applicant: CrowdStrike, Inc.

Inventor： Paul Sumedrea ,  Cristian Viorel Popa ,  Dragos Corlatescu ,  Vasile-Daniel Sava

IPC: G06F21/55

Abstract: An approach is provided that trains an artificial intelligence model (AIM) using training data to produce a generalized AIM, wherein the training data comprises log-collected data corresponding to multiple application types and the generalized AIM is trained to detect one or more cross-platform cybersecurity threats. The approach identifies multiple application-specific training data sets, wherein each one of the application-specific training data sets includes labeled application logs corresponding to one of the multiple application types. The approach then fine-tunes the generalized AIM using the multiple application-specific training data sets to produce multiple dedicated AIMs, wherein each one of the dedicated AIMs is trained to detect one or more application-centric cybersecurity threats targeted at a corresponding one of the application types.