Straiker Agentic AI Security Platform

Straiker, Inc. is an AI agent security company headquartered in Mountain View, California, with a team in Bengaluru, India. It was co-founded by CEO Ankur Shah, former SVP and GM of Prisma Cloud at Palo Alto Networks, and CTO Sreenath Kurupati, former VP of AI and security research at Akamai. It helps enterprises put AI agents to work safely as they gain access to tools, data, MCP servers and business workflows.

The platform has three products. Discover AI maps agents, MCP servers, skills and agentic workflows and monitors posture and misconfigurations. Ascend AI continuously red-teams agents for prompt injection, goal hijacking, tool misuse and inter-agent manipulation, using attack data from the STAR Labs research team. Defend AI detects and blocks remote code execution, data exfiltration, memory poisoning and destructive commands at runtime, in SaaS or on-premises, and an agent kill switch can take compromised agents offline. It covers coding, productivity and custom-built agents for sectors such as banking, healthcare and retail.

• Discover AI: Maps agents, MCP servers, skills and workflows with continuous posture and misconfiguration monitoring.
• Ascend AI: Continuously red-teams agents for prompt injection, goal hijacking, tool misuse and inter-agent manipulation.
• Defend AI: Detects and blocks code execution, file access violations, data exfiltration and destructive commands at runtime.
• Agentic kill switch: Cuts tool access, suspends sessions and takes rogue agents offline in seconds.
• Compliance evidence: Maps inventory, traces and logs to OWASP, MITRE ATLAS, NIST AI RMF and the EU AI Act.
• STAR Labs research: Feeds new attack findings into evaluations and detections.

A Vietnamese bank can red-team its customer chatbot and loan assistant before launch, then turn each finding into a runtime blocking policy. A hospital group can run Defend AI on premises to stop agents from exfiltrating patient data. An e-commerce company can use the kill switch to isolate a compromised support agent without stopping its other agents.

Every successful red-team attack from Ascend AI can be turned into a Defend AI runtime policy, closing the loop between testing and protection. A kill switch cuts tool access, suspends sessions or isolates one or many agents in seconds while preserving evidence. Agent inventory, traces, test history and policy logs are mapped to OWASP, MITRE ATLAS, NIST AI RMF and the EU AI Act for audits. On-premises deployment suits regulated and sovereign environments.

AI agents now execute code, call tools and move data, so security must go beyond filtering prompts to controlling actions. Straiker's own research reports that 36% of successful attacks on coding agents end in remote code execution. Banks and fintechs in Vietnam and Southeast Asia deploying customer-facing AI assistants need red teaming and runtime controls that can be shown to regulators.

Straiker, Inc. is located at 321 Castro St, Mountain View, California, with a team in Bengaluru, India. Its newsroom reports a 64 million USD Series A in June 2026, and in October 2026 it was listed as a Sample Vendor for Agentic AI Security in the Gartner Hype Cycle for Secure AI and Hype Cycle for AI-Driven Cybersecurity. Customer quotes include Omada Health, Coupa and EnterpriseDB.