OneCLI provides an open-source, self-hostable AI agent platform that gives every employee their own secure, autonomous agent with access to company tools and data. The platform combines a credential vault, policy enforcement layer, and per-agent micro-VM sandboxes to enable safe delegation of complex work. It is designed for teams that want to deploy AI agents across their organization without compromising security or control.
Funding
Funding not disclosed
Founders
Product
Problem
Companies struggle to deploy AI agents at scale because each agent requires individual setup of hosting, tool connections, and security credentials. Managing access control, approvals, and audit trails across multiple agents becomes an organizational burden, and giving autonomous processes direct API keys creates significant security risks. Existing solutions either lack the policy layer needed for enterprise use or force teams to build their own access management infrastructure.
Solution
OneCLI provides a complete AI agent platform where every employee gets their own persistent agent with dedicated memory, tools, and scheduled tasks, all managed through a single company workspace. The platform includes a gateway that stores credentials in a vault and attaches them to outbound requests at the proxy level, so agents never handle raw API keys. Each agent runs in its own micro-VM with a real browser and filesystem, providing strong isolation while enabling agents to perform complex tasks like running full application stacks or placing orders. The policy layer enforces approval rules outside the agent's control, ensuring that sensitive actions like sending emails or spending money always require human authorization. OneCLI is open source under Apache-2.0 and can be self-hosted on customer infrastructure or used through the hosted cloud service.
Target Audience
Primary customers are engineering and operations teams at companies that want to deploy AI agents across their organization while maintaining strict security, access control, and auditability. The platform serves both technical teams that need self-hosted deployment and non-technical employees who interact with agents through Slack or web interfaces.
Features
- Per-employee agents with individual identity, memory, and tool access, managed through a central workspace
- Micro-VM sandbox per agent with real browser, filesystem, and root access isolated from the host system
- Credential vault that proxies API keys and service credentials without exposing them to agents
- Policy engine that enforces approval workflows and access rules outside the agent's control
- Support for MCP tools, CLI commands, API calls, and custom scripts within the agent environment
- Slack and web interfaces for interacting with agents and receiving task results
- Audit trail tracking per employee, per agent, and per request for compliance and security review
- Offboarding capabilities that revoke all agent access and credentials when an employee leaves