Charlotte AI AgentWorks by CrowdStrike democratizes the creation of no-code security agents in Falcon

CrowdStrike has launched Charlotte AI AgentWorks, a no-code ecosystem that enables security teams to create, test, and deploy customized agents within the Falcon platform. Key partners include Accenture, AWS, Anthropic, Deloitte, Nvidia, OpenAI, Salesforce, and Telefónica Tech. The Charlotte Agentic SOAR feature orchestrates all CrowdStrike-native agents built through AgentWorks, as well as those from third parties, into a cohesive system.

For years, security teams in large organizations have grappled with a volume of alerts that far exceeds their analytical capacity. Security Operations Centers (SOCs) that have turned to conventional automation methods—such as rule-based correlation and first-generation SOAR scenarios—are struggling against the inflexibility of these systems in the face of ever-evolving attacker tactics. While detection capabilities are advancing, investigation and response efforts remain constrained by the number of available analysts.

AI agents offer a different solution to this challenge. These entities can assess variable contexts, chain together actions without explicit instructions at every step, and escalate to a human analyst when a decision exceeds their competency. CrowdStrike positions AgentWorks as the foundational layer for building and deploying these agents, making advanced capabilities accessible across its ecosystem. Without writing a single line of code, a security team can construct a detection, investigation, or response agent leveraging telemetry data from the Falcon platform, integrated language models (including Anthropic Claude, Nvidia Nemotron, OpenAI GPT), and Amazon infrastructure services like Bedrock and SageMaker.

Streamlining Security Agent Development

Charlotte AI AgentWorks empowers every Falcon subscriber to build agents tailored to their specific processes without the need for development skills. The platform provides connectors to Falcon telemetry data sources, incident response tools, and third-party integrations available in the CrowdStrike catalog. For example, a triage agent can access a terminal’s alert history, correlate it with known indicators of compromise, and trigger automatic network isolation based on thresholds defined by the security team.

The openness to language models from various providers offers flexibility suited to large organizations, some of which may already hold licenses and data processing agreements with specific model providers. The ability to select the underlying model without modifying the agent itself reduces substitute costs and mitigates the risk of vendor lock-in. Partnerships with Accenture, Deloitte, and Kroll introduce an industrial implementation dimension for organizations whose security teams may lack the maturity to deploy these agents independently.

Agentic SOAR: Orchestrating and Governing SOC-Level Operations

Charlotte Agentic SOAR serves as the orchestration layer overseeing all active agents within the CrowdStrike ecosystem. It unifies native platform agents, those built via AgentWorks, and trusted third-party agents into a coordinated system. The AgentWorks guardrails and process controls embedded in Agentic SOAR regulate interactions between agents and between analysts and agents, incorporating auditability mechanisms akin to those applied to human access in information systems.

For Chief Information Security Officers (CISOs) needing to justify the introduction of autonomous agents within their environments to executive management or regulators, this governance layer directly addresses the issue of decision traceability. Each action taken by an agent is logged, linked to the execution context, and subjected to the organization’s risk policies. CrowdStrike’s native AI Detection and Response integrates within this framework to identify abnormal behaviors even within the agent chain itself, adhering to an auditability logic that anticipates future governance obligations on AI systems expected from European regulators.

Cloud Act and Sovereignty: Unresolved Questions Persist

As a U.S. company, CrowdStrike is subject to the Cloud Act. The Amazon Bedrock and SageMaker infrastructures used to execute models within AgentWorks are operated by AWS, which is also under U.S. jurisdiction. For any European organization deploying AgentWorks agents that handle sensitive telemetry data—such as terminal logs and indicators of compromise for critical assets—this exposure necessitates explicit consideration in risk assessments compliant with NIS2 and DORA regulations.

None of the infrastructures mentioned in this launch have SecNumCloud certification or an equivalent qualified European certification. While this absence does not impede deployment for most private organizations, it complicates usage in highly regulated sectors (healthcare, defense, energy, critical infrastructure) and imposes specific contractual clauses for entities obligated to adhere to data localization requirements. CISOs considering AgentWorks will need to evaluate the categorization of data processed by each agent prior to any production deployment.

Scroll to Top