Governing AI agents in production: what's new in the Redpanda Agentic Data Plane

The Redpanda Agentic Data Plane (ADP) provides agentic governance, connectivity, and context at enterprise scale. We’re pleased to announce that it has reached general availability (GA) status and is deployable on Amazon Web Services (AWS). 

Redpanda Agentic Data Plane runs in your account, using your existing discounts and credits. It's already in production and ready for you to adopt within your own organizations, so you can run agents on all your data, safely.

‍

Why AI agent governance is so hard

The gap today is getting agentic applications working across departments or the whole organization. Too many stay stuck at the prototype or MVP stage, needing a path to production and secure, reliable integration with operational systems.

Agentic apps also demand flexibility. AI innovation moves so fast that no one wants monolithic lock-in. No single cloud, frontier LLM provider, framework, or data platform can solve the entire stack. Plus, customers want to iterate quickly, integrate freely, and compose their architectures creatively. No walled gardens. 

However, today's agentic tools are fragmented: separate systems for governance, spend management, observability, runtimes, and MCP servers. Enterprises are forced to cobble together stacks or build internally, significantly delaying time to value.

The Redpanda Agentic Data Plane solves this. 

It's a comprehensive, enterprise-ready, fully managed service that unifies all your AI models, agents, and MCP-connected data sources into a single data plane with built-in governance: policy management, guardrails, kill switches, and budget controls. 

Organizations stay nimble and mission-focused on critical code and business deliverables, not infrastructure. While executives gain full visibility into enterprise-wide agent adoption, complete audit transcripts, policy enforcement, and budget.

What we’re delivering

Redpanda Agentic Data Plane builds on our core platform and delivers a critical set of features for enterprise agentic AI adoption.

• MCP server management. Centrally control all MCP servers, which agentic systems need to access mission-critical data. Redpanda Agentic Data Plane includes a library of highly optimized, token efficient MCP servers for popular systems (SQL databases, Salesforce, ServiceNow, Workday, Google Drive, Slack, Jira, and more), as well as no-code workflows to build new ones, and registration of existing servers for a single unified view.
  
  
• Agent registry and BYO Agents. A single pane of glass for managing agents. Whether built into the Redpanda Agentic Data Plane or elsewhere, every agent becomes a discoverable resource that others can use and develop, closing a major governance gap.
  
  
• Transcripts and governance dashboard. Every prompt, response, tool call, and agent action is recorded in audit-grade, immutable OpenTelemetry (OTel) format with an interactive timeline. The dashboard turns this into instant answers: usage and cost per provider, model, and user, so leadership can see in seconds what's running and what's driving spend.
  
  
• AI gateway and guardrails. Register all LLM providers for secure OIDC proxy access, eliminating credential sprawl and shadow AI. Centrally configure policies and filters across all providers, with violations tracked in transcripts rather than at the application level.
  
  
• Token vault and OAuth providers. Integrates with your corporate IdP via OIDC for "on-behalf-of" authorization, so agents are clearly distinguished from human users and granted least-privileged access. No more handing credentials to agents.
  
  
• Microsoft Teams integration. Brings managed agents directly into everyday workplace applications, lowering adoption barriers.

Get in touch to learn more

Redpanda Agentic Data Plane is delivering new capabilities and features almost every day. Richer transcript analysis, ever more sophisticated guardrails, topology graphs and violations handling, more robust capabilities to Bring Your Own Agents (BYO Agents), spend alerts, LLM provider routing, and more. 

If you’re deploying agents at scale in your enterprise, we’d love to help. Get in touch for a demo or to discuss your agentic AI goals.