Charlotte AI is CrowdStrike’s agentic artificial intelligence (AI) security analyst, built into the Falcon platform. You ask it questions in plain language, it assesses detections, coordinates specialized agents across domains during an investigation and runs the agentic workflows your security teams design. Response actions stay off until your team configures them and decides which ones need human approval.
What you’ll take from this
This guide is for security professionals and managers deciding how far to trust AI with company data. Founderz’s Responsible AI Leadership program covers the governance side of that decision.
- Understand what the tool does today and how it changed since launch.
- Follow a case from the first signal to a verdict, step by step.
- Check the data protection controls before you switch on automation.
- Know where the free tier ends and where a person still decides.
What is Charlotte AI?
It is a multi-agent AI system that runs natively on the CrowdStrike Falcon platform, and CrowdStrike describes it as neither a chatbot nor a single agent. One layer handles questions, while others do command-line analysis, exposure analysis, query writing, workflow generation and threat intelligence work.
CrowdStrike introduced it on May 30, 2023 as a “generative AI security analyst” and made it generally available on February 20, 2024. It now anchors the vendor’s agentic SOC products. In November 2025 it received FedRAMP High authorization for US federal agencies.
How does the agentic security workforce work on the Falcon platform?
Its agents use the telemetry Falcon already collects to unify endpoint, identity, cloud workload and network signals. They split an investigation into parallel tasks and return one verdict with its reasoning. A typical case moves through five steps.
- A detection fires, or someone types a prompt such as “Show me critical vulnerabilities on internet-facing endpoints”.
- Agents assemble cross-domain context: the host and user involved, any known vulnerability with its CVE (Common Vulnerabilities and Exposures) entry and intelligence on the adversary or threat actor. A zero-day vulnerability has no CVE yet, so behavior carries more weight.
- The scoring agent returns a verdict, a confidence score, a recommendation and an explanation.
- Your rules decide whether the case is escalated, closed or routed to a person.
- Any remediation runs only if your team has configured it, with approval where you require it.
CrowdStrike benchmarks that agent against decisions made by its own Falcon Complete managed detection and response team. Test it on a sample of your own cases before it closes anything unreviewed.
What can security teams orchestrate, from one workflow to custom agents?
Most teams start with questions and first-pass review, then move to custom agents once they trust the output.
| Capability | What it does | Who decides |
|---|---|---|
| Conversational queries | Lets you interact with the Falcon platform through questions and writes the search query for you | The person who asks |
| Detection Triage | Scores first-party detections with a verdict, confidence level and explanation | Your SOC lead |
| Prebuilt agents | Handle vulnerability and exposure prioritization, malware analysis, reconnaissance and query writing, returning actionable findings | The team that reviews their output |
| Charlotte AI AgentWorks | Builds custom agents with no code, on Anthropic Claude, OpenAI GPT or NVIDIA Nemotron models | The agent owner, through role-based permissions |
| Charlotte Agentic SOAR | Coordinates agents and fixed workflows, and connects third-party tools through MCP | Whoever approves each automated action |
AgentWorks lets you build and customize agents around your team’s own knowledge. Incident response playbooks or escalation rules become an agent that works around the clock at machine speed. See also AI tools for cybersecurity teams.
How does it protect your data and access rights?
It works inside the permissions each user already has, so nobody sees data through it that they are not cleared to see. Every answer traces back to the data behind it, and every action is bounded, user-authorized and logged, according to the vendor’s product documentation.
The vendor also states that the product is certified under ISO/IEC 42001:2023. That is the international standard for AI management systems. Controls include execution traces, agent version history with rollback, credit caps and approval steps. Before you connect sensitive systems, agree on four points with your data protection officer.
- List which data sources each agent can read, including third-party and SaaS data sent into Falcon.
- Map Falcon roles so that agent permissions match human ones.
- Decide which actions run on their own and which wait for sign-off.
- Review audit logs and execution traces on a fixed schedule.
In 80% of incidents Microsoft investigated, attackers aimed to steal data, according to the Microsoft Digital Defense Report 2025. The same report says phishing-resistant multifactor authentication can block over 99% of identity-based attacks. AI-powered detection helps you stop an intrusion before it becomes a breach; it does not replace those basics.
Why are security operations teams using AI to accelerate alert handling in 2026?
CrowdStrike argues that AI-driven adversaries now move faster than manual review. Security leaders expect AI to shape the next year of computer security across the whole attack surface, from cloud computing to identity.
Among respondents to the World Economic Forum’s Global Cybersecurity Outlook 2026, 94% expect AI to be the most significant driver of change in the field in the year ahead.
In the same WEF report, 87% named AI-related vulnerabilities the fastest-growing cyber risk of 2025. The share assessing the security of their AI tools rose from 37% to 64%.
The NIST Cyber AI Profile (NIST IR 8596), a December 2025 preliminary draft, splits the work into three areas: securing AI systems, AI-enabled cyber defense and thwarting AI-enabled attacks. Agentic AI tools like Charlotte sit in the second group, so the first still applies to them.
Is Charlotte AI free?
Partly. Charlotte AI’s free tier lets qualifying CrowdStrike customers use it at no cost from the Falcon console, with 50 AI credits that renew each month. Beyond that, CrowdStrike sells credit-based Agentic SOAR plans on request. The detection-scoring and response agents sit in the fuller tier, so check current terms before you plan a rollout.
Where agents help and where a SOC analyst still decides
An AI agent is strongest on repetitive, high-volume work and weakest where the context lives outside the platform.
- Hand them enrichment, query writing, malicious script explanations and first drafts of incident reports, and let them prioritize the queue.
- Keep people on containment that could stop a business service, legal notifications and insider cases.
Before you deploy a new agent widely, run it in recommendation-only mode for a few weeks. Compare its verdicts with your team’s and widen its permissions one action at a time. Our guide to threat hunting covers what experienced human hunters still find that automated review tends to miss.
Further reading
These primary sources cover the company and the standards.
- CrowdStrike (Wikipedia)
- Cyber AI Profile, NIST IR 8596 (NIST, 2025)
- Security orchestration, automation and response (Wikipedia)
Frequently asked questions
Short answers to the questions security teams ask most often before they switch it on.
Can Charlotte AI take response actions on its own?
Not by default. Prebuilt agents only produce information, summaries and recommendations. Actions that change your environment must be configured in Agentic SOAR workflows or AgentWorks agents. You choose whether each one runs autonomously or waits for human approval.
Do you need to code to build agents in AgentWorks?
No. You describe the agent’s mission in everyday words, add knowledge files, assign authorized tools and set the conditions that trigger it. Execution traces, audit logs and credit caps come built in.
Is CrowdStrike an AI company?
CrowdStrike is a security software company, listed on Nasdaq as CRWD, whose Falcon products use AI to stop breaches through endpoint security and threat detection. Its agentic analyst is one product on that platform, alongside Next-Gen SIEM and exposure management.
Choosing what a team may automate, and who signs off, is a governance skill. Founderz’s Responsible AI Leadership program teaches professionals to assess AI risks, set transparency rules and lead responsible AI adoption across their organization.
