Run agents on all your data, safely.
An AI Gateway, MCP servers, declarative agents, and built-in governance: one platform so your team can move from prototype to production without rebuilding the foundations every time.
Pick where to start
Build a new agent
Start from scratch with governance built in. Set up an agent, connect your data, and deploy with identity, routing, and logging from day one.
Get startedBring an agent you already built
Get visibility and control over agents built elsewhere. Same gateway, same dashboard, same identity and access controls, without rebuilding.
Set up your agentPopular across Agentic Data Plane
One gateway for every LLM call
Route all agent traffic through a single point. Set budgets, switch providers, and enforce policies without changing agent code.
Governed MCP servers for your data
Expose databases, APIs, and files as tools your agents can use with access control and audit logging built in.
One dashboard for all activity
See every agent call, token spent, and tool invocation. Debug issues, track costs, and prove compliance from a single view.
Identity on every call
Every call runs as a user or as an agent's own identity, and user-delegated OAuth carries each person's identity to the tools that need it. Know who did what, and enforce permissions at the data layer.
Build in Agentic Data Plane versus bring your own
When to use declarative agents versus registering agents you built elsewhere.
How agents, tools, and data fit
The parts of a declarative agent and how it reaches tools and data.
Need a hand?
Ask Redpanda AI, search the docs, or talk to a real human in Slack.