Conduit FAQ

Yes. Conduit gives you one layer to govern your AI agents, Mantle's and any other vendor's. It manages their identity, policy, audit and cost in one place.

Yes. Conduit is built for companies like yours, where each department buys its own agents. You know, scope, audit and cost each agent, whichever vendor built it.

Yes. You keep your choice of vendors. Conduit is vendor-neutral by design. It turns a sprawl of agent endpoints into one controlled, observable layer, without locking you into any single vendor.

Yes. You decide what each agent can call, for whom and within what budget, through one policy layer. Each agent gets its own verifiable identity, scoped to what it may do, with short-lived credentials. Each decision is logged with its reason.

Yes. You see what each agent did and what it cost, in one place. Conduit keeps traces of prompts, tool calls and model responses, with replay. One cost ledger covers model spend, agent runtime and tool calls, split by team, agent and customer. You can export audit data to your security monitoring (SIEM) tools.

Yes. Conduit places prompt-injection defense at the boundary between the agent and your tools, not inside the agent. The check sits where agents reach your systems, whichever vendor built them, and each policy decision is logged with its reason.

Yes. Conduit is vendor-neutral, so you keep the agents and models you have. It works with agents from other vendors, the major model providers and your in-house models. It supports the open MCP and A2A agent protocols, so you expose each system once and any agent that needs it can use it.

Yes. Your credentials stay in one place. Conduit keeps them in a central vault and issues short-lived credentials, so no agent stores long-lived tokens. Each agent signs in machine to machine with its own workload identity, using standard OAuth 2.0 token exchange.

Yes. Your model traffic goes through one gateway. It works across the major model providers and your in-house models. It routes, caches, rate-limits, retries and falls back, so you manage model calls in one place.

Yes. Your agents reach your SaaS apps with a scope you can audit. Conduit is compatible with Cross-App Access (XAA), and each system is exposed once, through its own MCP server, to the agents that need it.

Yes. Conduit is built for companies where each department buys its own agents. It governs Mantle's agents and any other vendor's from one layer. To begin, use the form on our contact page, and our team will be in touch.

Yes. Use the form on our contact page, and our team will be in touch. Pricing depends on scope.

Yes. You decide the budget. Conduit's policy plane sets rules per agent, per user, per tool and per budget, and each decision is logged with its reason. One cost ledger shows the spend by team, agent and customer.