Pillar Security

Pillar Security is an end-to-end platform for securing AI applications and agents. It builds a live AI-BOM mapping agents, models, prompts and MCP tools, runs autonomous adversarial testing via its RedGraph agent, and enforces adaptive, model-agnostic runtime guardrails. Founded in 2023 and backed by a $9M seed round, it is independent and early-stage but carries a strong signal as a 2026 Gartner Cool Vendor in AI software security.

Pillar Security provides a platform for securing AI applications and agents across their full lifecycle. It builds a live AI Bill of Materials (AI-BOM) that maps every AI asset -- agents, models, prompts, MCP tools, skills, datasets and configurations -- from source code, endpoints and CI/CD. This discovery layer gives security teams visibility into the sprawling and fast-changing surface introduced by generative and agentic AI. On top of inventory, Pillar offers two headline capabilities. RedGraph is an autonomous adversarial agent that maps the attack surface and runs multi-step, dynamic attack scenarios -- chaining tool calls and performing reconnaissance the way a real attacker would -- to find weaknesses in agents and their tools, MCP servers and permissions. Complementing this, Pillar's runtime guardrails are model-agnostic and application-centric, evolving from telemetry, red-teaming insights and threat intelligence rather than fixed rules, and enforcing approved-model lists, MCP allowlists and usage policies. Pillar was founded in October 2023 by Dor Sarig and Ziv Karliner and operates from Tel Aviv and New York. It is independent and venture-backed, having raised a $9M seed round led by Shield Capital (with Golden Ventures, Ground Up Ventures and angels) in April 2025. It was recognized by Gartner as a 2026 Cool Vendor in AI software security, citing its RedGraph offensive testing, dynamic runtime guardrails and SAIL framework. Buyers should weigh that positive analyst signal against the reality that it is still an early-stage company.