Kiteworks launches agent marketplace for governed AI agents

Kiteworks launched the Kiteworks Agent Marketplace on September 16, 2026, giving customers more than 60 AI agents that run on the same control plane used for regulated human-data interactions. The company says the marketplace lets organizations deploy agentic AI quickly without building governance, security and compliance into each agent first.

Why it matters: - Kiteworks is trying to remove a key barrier to enterprise AI adoption: security and compliance work that usually has to be rebuilt for every new agent. - The marketplace is designed to let customers deploy governed AI agents without creating a separate control stack for each one. - The launch expands Kiteworks’ compliance model from human users to AI agents that access regulated data.

What happened: - Kiteworks announced the Kiteworks Agent Marketplace on September 16, 2026. - The marketplace includes more than 60 ready-to-use AI agents. - The agents install directly in Claude and connect to a customer’s Kiteworks environment through the Kiteworks Secure MCP Server. - Support for additional AI platforms, including ChatGPT and Microsoft Copilot, is expected soon. - The marketplace is available now.

The details: - The Marketplace currently offers more than 50 agents and publishes new agents and updates automatically. - The agent set includes compliance checks for GDPR, CCPA, HIPAA, CMMC, NIST CSF, NIST SP 800-53, ISO 27001, ISO 42001, SOC 2, PCI DSS, FedRAMP, DORA, NIS2, ITAR, EAR and other frameworks. - It also includes accessibility checks for WCAG and Section 508. - Additional agents cover data governance and productivity tasks such as document summarization, contract review, retention and release audits. - The marketplace is available in English, German, Dutch, Hebrew and Japanese. - Each agent connects through the Kiteworks Secure MCP Server. - The server extends role-based and attribute-based access controls, encryption and audit logging from human users to agent interactions. - Credentials never enter the AI model context. - Every agent action is tied to the authenticated user who approved it and is recorded in the same audit trail as human activity. - Kiteworks calls this approach Compliant AI. - Agent and skill logic is published as open, human-readable definitions. - Customers can review what an agent does before running it, request changes or make changes themselves. - The marketplace is live at the Kiteworks Agent Marketplace.

Between the lines: - Kiteworks is positioning governance as a product feature, not a customer implementation project. - The open, inspectable agent definitions are meant to build trust and speed iteration without waiting for a standard release cycle. - The company is extending its existing data-control model into an AI-agent layer rather than treating agents as a separate system.

What’s next: - Kiteworks plans to add support for more AI platforms, including Microsoft Copilot, ChatGPT, Grok and others. - The company also plans to keep releasing improvements agent by agent through customer feedback.

The bottom line: - Kiteworks is betting enterprises want agentic AI, but only if the agents inherit the same governance, auditability and access controls already used for sensitive data.