Barndoor AI

What Barndoor AI does

Barndoor AI provides a centralized control plane for managing Agentic AI, offering visibility, control, and guardrails across LLMs, agents, MCP servers, tool calls, and data. It enables enterprises to scale AI safely by providing granular access, policy enforcement, and oversight. The platform helps manage AI usage, MCP servers, and tool calls, empowers users with approved AI, enforces least-privilege permissions, and audits every agent action. It also includes an LLM Gateway to route calls to models, manage access, control costs, and protect sensitive data in transit. Barndoor aims to provide a governance layer for running AI in production safely, visibly, and at scale, without vendor lock-in.

Products

• MCP GovernanceConnects every AI agent and tool, providing access controls and an audit trail. It manages MCP connections, enables intelligent MCP tool selection, enforces least-privilege permissions for AI, and audits every agent action across the stack. Includes features like RBAC & ABAC policies, MCP tool orchestration (ToolIQ), and audit logs for every AI action.
• LLM GatewayRoutes every call to the right model, manages access, and controls cost with full visibility. It connects every model through one endpoint, routes to the right one, and fails over when providers go down. It also protects sensitive data from reaching external models, controls AI cost across teams, and tracks model performance.

Key capabilities

• Visibility, controls, and guardrails across every LLM, agent, MCP server, tool call, and data
• Connect every AI agent and tool, safely
• Manage every agent and MCP connection
• Access controls and audit trail
• See and control AI usage, MCP servers, and tool calls
• Empower users with approved AI and MCPs
• Enable AI with intelligent MCP tool selection
• Enforce least-privilege permissions for AI
• Protect sensitive data before it flows through a tool call
• Audit every agent action across your stack
• Route every call to the right model
• Manage access and control cost with full visibility
• Connect every model through one endpoint
• Fail over when providers go down
• Keep sensitive data out of your models
• Protect every prompt and response in flight
• Stop PII, secrets, and IP from ever reaching an external model
• Control AI cost across teams
• Set budgets, rate limits, and token caps by user, team, or app
• Track latency, errors, tokens, and cost across every model
• No AI vendor lock-in
• Connect to any MCP (Official, custom, local, or Barndoor-hosted)
• Predictable AI spend with caps
• Universal kill switch
• Deploy Barndoor as SaaS, Private Cloud, or On-Prem
• RBAC & ABAC policies
• MCP tool orchestration (ToolIQ)
• Audit logs for every AI action
• AI & MCP registry
• Unlimited MCP servers
• Unlimited AI & agents connections
• Automatic MCP tool discovery
• Identity provider (IdP) integration
• Runtime policy enforcement
• Machine-to-machine authorization
• Platform APIs (policy as code)
• PII & credential detection
• Pre-built & custom DLP classifiers
• Indirect prompt injection detection
• Traces
• Usage dashboard
• Policy change history
• Cost control
• Prompt security
• Universal API
• Unlimited models
• Credential management & virtual keys
• Model access with RBAC
• Runtime access control
• Model routing
• Fallbacks & health checks
• Token & cost budgeting
• Rate limiting
• Direct prompt injection detection
• Air-gapped deployment option
• Field-level data protection and controls
• Response transformation for MCP tool traffic
• New MCP connectors (Klaviyo, Statsig, Google BigQuery, Cloudflare API, Cloudflare DNS Analytics, Apify)
• SCIM provisioning
• Python SDK
• Typescript SDK
• API for programmatic management of MCP governance and LLM gateway

Use cases

• Scaling AI safely across the enterprise
• Managing AI usage and MCP servers
• Empowering users with approved AI and MCPs
• Enforcing least-privilege permissions for AI
• Auditing every agent action
• Routing calls to the right LLM model
• Managing access to LLM models
• Controlling costs of LLM usage
• Protecting sensitive data in LLM prompts and responses
• Tracking LLM model performance
• Ensuring compliance with AI governance policies
• Preventing AI vendor lock-in
• Controlling AI spend across teams and users
• Quickly disabling AI access with a universal kill switch
• Deploying AI governance in various environments (SaaS, Private Cloud, On-Prem)
• Setting budgets on expensive models and falling back to cheaper ones
• Routing to alternative LLM models if one hits a rate limit or becomes unavailable
• Controlling access to premium models based on limits
• Stripping specific fields from MCP tool responses before they reach the agent
• Removing specific fields from MCP tool requests before they are sent to the underlying system
• Connecting personal accounts to MCP servers for end-users
• Integrating with marketing automation platforms (e.g., Klaviyo) for agents to pull customer segments or trigger actions
• Integrating with product development platforms (e.g., Statsig) for agents to check feature flags or manage rollouts
• Integrating with data warehouses (e.g., Google BigQuery) for agents to query tables or pull analytics
• Integrating with cloud infrastructure APIs (e.g., Cloudflare API) for agents to manage account functions
• Integrating with DNS analytics (e.g., Cloudflare DNS Analytics) for agents to gain visibility into DNS performance
• Integrating with web scraping platforms (e.g., Apify) for agents to launch scraping jobs and retrieve data

AI approach

Barndoor AI provides an agentic governance platform for AI, focusing on visibility, control, and guardrails for LLMs, agents, MCP servers, tool calls, and data. It enables secure scaling of AI by managing access, enforcing policies, and auditing actions across various AI components. The platform includes an LLM Gateway for routing calls, managing access, controlling costs, and protecting sensitive data, and MCP Governance for connecting and controlling AI agents and tools.

Tech named: LLM Gateway, MCP Governance, ToolIQ

What it says sets it apart

• Centralized control plane purpose-built for Agentic AI
• Intelligent and continuous governance, not just reactive
• Comprehensive visibility, controls, and guardrails across all AI components
• Ability to connect any model provider and MCP without vendor lock-in
• Predictable AI spend with caps and full visibility
• Universal kill switch for immediate control over AI operations
• Flexible deployment options: SaaS, Private Cloud, On-Prem, Air-gapped
• Field-level data protection and response transformation for MCP tool traffic
• Extensive set of MCP connectors for various domains (marketing, data, infrastructure)
• Robust security philosophy: Least Privilege, Defense-in-Depth, Consistent Implementation, Continuous Improvement
• SOC 2 Type II Attestation
• Founding team with experience in secure API Management and enterprise AI adoption

Funding rounds we track

From the AI funding tracker — rounds as reported by the linked publications.