Redpanda Introduces AI Gateway to Bring Control and Governance to Enterprise AI
New Redpanda Agentic Data Plane capabilities extend a unified governance layer for AI agents, MCP servers, and real-time enterprise data, enabling agentic safety and trust
SAN FRANCISCO, Feb. 18, 2026 — Redpanda today announced new Redpanda Agentic Data Plane (ADP) capabilities designed to help enterprises move agentic AI from experiments into production with stronger control, visibility, and governance. The release introduces an AI Gateway along with end-to-end observability and evaluation (via OpenTelemetry), agent interoperability, and unified authentication and authorization—creating a central governance layer for connecting AI agents and Model Context Protocol (MCP) servers to live, real-time enterprise data.
As more organizations deploy agents that can access sensitive systems and data, the operational challenge is increasingly about governing agent behavior—who (or what) can connect, what they can do, and how actions are tracked. Redpanda ADP is positioned as the “connectivity control point” for agent interactions, enabling policy enforcement and auditability without requiring teams to implement bespoke controls across every downstream data source.
What’s included in this ADP update
- AI Gateway (central AI traffic control): A unified access layer for applications, models, and MCP services—built to centralize routing, policy enforcement, usage and cost controls (e.g., budgets/limits), and operational visibility across agent workflows.
- MCP governance and administration: An admin-managed approach to aggregating and controlling MCP servers, including configuration-driven deployment to standardize what’s approved and how it’s used.
- Agent flexibility and interoperability: Support for agents built with existing frameworks, plus optional built-in Redpanda agents—working through open standards and integrating with ADP services for identity, authorization, and observability.
- Unified identity and fine-grained access control: OIDC-based authentication and granular authorization policies to reduce reliance on long-lived credentials and better contain agent access.
- Observability + evaluation using OpenTelemetry: Full-fidelity telemetry (metrics, traces, logs, and transcripts) that can be reviewed in the Redpanda console or exported to external tools for debugging, compliance, and post-incident analysis.
Redpanda also highlighted its real-time foundation—supporting continuous context updates and event-driven execution—along with connectivity via Redpanda Connect and its large catalog of connectors for integrating data sources without re-architecting existing systems.
Availability
These ADP capabilities are available today to select Redpanda Design Partners running Bring Your Own Cloud (BYOC) on AWS, with GCP, Azure, and self-hosted support planned for future releases.
Read the full announcement here.
Real-World Impact,
Proven Results
Explore Redpanda solutions—from financial services and telecommunications to real-time analytics and event-driven architectures. Discover how Redpanda can help you build faster, more resilient, and cost-efficient streaming pipelines for your specific challenges.