Enterprise AI Security and Compliance Platform

COMPANY OVERVIEW

Enkrypt AI, founded in 2022 by Yale PhDs Sahil Agarwal (CEO) and Prashanth Harshangi (CTO), is a Boston-based company that provides a security and governance layer for enterprise generative AI deployments. The company raised a $2.35 million seed round in February 2024, led by Boldcap with participation from Berkeley SkyDeck, Kubera VC, Arka VC, Veredas Partners, and Builders Fund. Its platform, Enkrypt AI Sentry, acts as a control layer between users and large language models, detecting vulnerabilities, reducing jailbreak rates, and enforcing security, privacy, and compliance policies so enterprises can safely adopt LLMs across finance, healthcare, and other regulated sectors.

CORE FOCUS

Enterprise adoption of generative AI introduces security risks that traditional controls were not designed to handle — prompt injection, jailbreaks, data leakage, and policy violations that occur inside LLM interactions rather than at the network perimeter. Enkrypt AI addresses this by sitting between users and AI models as an inline control layer, enforcing policies in real time without requiring changes to underlying AI infrastructure. The platform converts human-readable compliance policies into automated guardrail rules, eliminating the manual configuration burden that has slowed enterprise AI adoption. Automated red teaming continuously probes deployed AI systems for vulnerabilities before adversaries can exploit them, giving security teams proactive visibility into AI risk posture.

PRODUCTS & TOOLS

Enkrypt AI Sentry — Inline security and governance platform that enforces policies between users and LLMs in real time.

• Intercepts and inspects every interaction between users and AI model endpoints
• Blocks unsafe outputs, policy violations, and data leakage with full audit logs
• Provides real-time explanations for blocked or flagged interactions
• Supports multiple model providers and custom authentication methods

Policy Engine — Automated policy-to-rule conversion that achieves compliance at 100x the speed of manual configuration.

• Converts human-readable compliance policies into automated guardrail rules
• Auto-generates and allows editing of custom rules without engineering effort
• Maps organizational policies directly to LLM behavior enforcement

Automated Red Teaming — Continuous adversarial testing that identifies AI vulnerabilities before deployment and in production.

• Runs automated attack tests based on use-case-specific threat scenarios
• Produces detailed attack logs and risk score overviews for security teams
• Validates model resilience against prompt injection, jailbreaks, and CBRN misuse

Endpoint Management — Centralized control of all AI model endpoints across the enterprise.

• Add and configure any model endpoint through a unified interface
• Supports providers including OpenAI, Anthropic, Google, Azure, and custom deployments
• Applies consistent governance policies across all connected model endpoints