Noma Agent Security and Governance Platform

Noma Security is an AI and agent security company led by co-founder and CEO Niv Braun. It came out of stealth in November 2024, issues its press releases from New York City and runs its product, R&D and research teams in Tel Aviv. It helps enterprise security teams keep control as employees, engineers and business teams adopt AI agents that drift, get manipulated or reach sensitive systems.

The platform combines four products that share context. AI-SPM discovers every agent, MCP server, skill and model and ranks their risks; Agent Access Control sets what each agent may do by user, team or tool sensitivity; AI-DR weighs each agent action against the full session and can block, steer, mask data, alert a human or trigger a kill switch; and AI Red Teaming runs multi-turn attacks against AI apps. It covers endpoint agents, cloud agents built on AWS Bedrock, Azure AI Foundry, Databricks, LangChain or CrewAI, and SaaS agents on Copilot Studio, Salesforce Agentforce and ServiceNow, enforcing policy through existing gateways and agent hooks.

• AI-SPM: Discovers every agent, MCP server, skill and model and prioritises their risks.
• Agent access control: Approves, reviews or blocks agents, MCP servers and skills and scopes their actions.
• AI-DR: Detects and stops risky agent behaviour at runtime with session context.
• AI red teaming: Tests AI apps with multi-turn attacks before release.
• AI supply chain security: Scans models, MCP servers, tools and packages and assigns a risk status.
• AI compliance: Produces inventory, framework mappings, test results and audit trails.

A Vietnamese bank can inventory agents built by business units in Copilot Studio and block those that can move customer data outside approved systems. A retailer running a customer-facing assistant on AWS Bedrock can red-team it and then enforce runtime policies. A technology company can govern which MCP servers and skills developers connect to coding agents.

Security teams can find agents nobody reported; the vendor says enterprises routinely discover 10 to 100 times more agents than expected on SaaS platforms. Runtime context lets agents stay connected to sensitive systems while risky combinations of legitimate actions are stopped. Policy is enforced through existing gateways and agent hooks, so no re-architecture is needed. Framework mappings, test results and audit trails give evidence for board and regulator questions on AI risk.

Agent adoption is spreading from developers to business teams that build agents in Copilot Studio, Agentforce and ServiceNow without telling security. Gartner has created an AI application security category with agentic runtime security at its core. Vietnamese and Southeast Asian enterprises building on AWS, Azure and Databricks need the same inventory and runtime controls as AI regulation tightens across the region.

Noma Security came out of stealth in November 2024 and in July 2025 raised a 100 million USD Series B led by Evolution Equity Partners with Ballistic Ventures and Glilot Capital, reporting ARR growth above 1,300% in a year. Other backers include Cyber Club London, Databricks Ventures and SVCI. It says it is trusted by over 10% of the Fortune 100 and was named a Market Shaper by Gartner in 2026.