Knostic vs Prediction Guard

Knostic is an AI security company founded in 2023 by Gadi Evron (CEO) and Sounil Yu (CTO). It is based in Herndon, Virginia. It started out selling need-to-know access controls for enterprise AI assistants such as Microsoft 365 Copilot and Glean, which keep answers from exposing data a user shouldn't see. Its website now calls it "the control plane for agentic AI on the endpoint", and it sells two products for AI coding assistants and agents. Kirin runs alongside Cursor, Claude Code, GitHub Copilot, Windsurf, Codex, JetBrains, Devin Desktop and Gemini CLI, and supports Claude Cowork through a plugin. It checks each tool call against policy before it runs and blocks it, downgrades it or sends it to a human. It also validates MCP connections, scans dependencies for hallucinated, vulnerable or malicious packages, and finds and redacts secrets before they reach the model. It audits agent instruction files such as CLAUDE.md and .cursorrules. Kirin installs as a lightweight user-space client that admins can roll out with MDM scripts, on macOS and Windows; Linux is in beta. Its Shadow AI Discovery reports unsanctioned AI use but doesn't block it. AgentMesh scans agent skills, MCP servers and VS Code extensions and labels each one dangerous, risky or clean. In June 2026 VirusTotal added AgentMesh analysis of VS Code extensions. Knostic also publishes an open-source scanner, OpenAnt. Knostic still has a page that pairs it with Microsoft Purview against Copilot oversharing, and its AWS Marketplace listing is still the Copilot Readiness product. Its homepage, product and pricing pages no longer describe that product. It raised a $3.3 million pre-seed round in April 2024 and an $11 million round led by Bright Pixel Capital in March 2025. In 2025 it also received a $5 million investment as an RSAC Innovation Sandbox finalist. In June 2025 Virginia Business put its total funding at $19 million and its headcount at about 30. In October 2025 Gartner named Knostic a Cool Vendor in AI Cybersecurity Governance, for its controls on AI oversharing. **AI compliance coverage** Knostic doesn't map its products to the EU AI Act, the NIST AI RMF or ISO 42001. Its trust center lists a SOC 2 Type 2 report and GDPR, its FAQ offers a compliance package with the SOC 2 report and pen test results, and its Enterprise plan includes compliance reporting. **Who it's for** Security teams at companies whose developers use AI coding agents, agent skills and MCP servers. It is not a governance, model inventory or model testing platform.

Prediction Guard sells a self-hosted "sovereign AI control plane" for regulated organizations. It sits between applications and the models, agents and MCP tools they call, and applies access, safety and policy controls to each request inside the customer's own infrastructure. The company was founded in 2023 in Lafayette, Indiana; CEO Daniel Whitenack is a co-founder. It raised a $3.7 million seed round led by Sovereign Capital in May 2025, and Noblis Ventures made a strategic investment of undisclosed size in July 2025. At the time of the seed round it had eight employees and said it had 18 customers in health care, defense, transportation and manufacturing. The platform registers each agent with an identity and role and limits it to the models and tools that role needs. Runtime controls include PII masking, prompt injection detection, toxicity filtering, input sanitization, MCP tool scoping and human-in-the-loop approval; requests that fail a policy are blocked and logged. It produces an exportable AI bill of materials (CycloneDX) of models, guardrails and MCP servers, runs model security scans and streams security events to SIEM tools such as Splunk, Datadog and Grafana. Applications call it through OpenAI- and Anthropic-compatible APIs. It deploys into a customer VPC on AWS, Azure or Google Cloud, on Kubernetes on premises, or air-gapped. Prediction Guard says its control overhead is under 200 ms per request. **AI compliance coverage** Prediction Guard says its policies align with the NIST AI RMF, NIST AI 600-1, the OWASP Top 10 for LLM and agentic applications, OMB M-26-04 and AIUC-1, with one-click baselines and an exportable "proof of alignment" for the NIST AI RMF. A company blog post (August 2026) adds the EU AI Act and ISO/IEC 42001. The same post labels its OWASP mapping "self-published" and not independently verified, and notes that the audit log records enforcement events rather than full replayable traces. We found no published control-by-control mapping. The homepage shows badges for a SOC 2 Type 2 report by Sensiba and HIPAA compliance; there is no public trust center. **Who it's for** Regulated enterprises and public sector bodies that want AI guardrails and agent controls to run inside their own network, including disconnected environments. It is sold to government through Carahsoft. Its homepage quotes executives at Noblis (also an investor), SimWerx and iBase-t.