Sep 10One Agent, Three Surfaces

The Three Buckets of the Agentic World

Agentic software splits into three buckets: agent orchestration (how agents reason, call tools, and coordinate), agent-user connectivity (how agents plug into user-facing applications), and agent learning (how deployed agents improve from every interaction). Most of the industry lives in the first bucket; most of the user-facing value lives in the second and third.

Orchestration

Agent-user connectivity

Learning in production

The three buckets: agents acting, agents meeting users, agents improving.

The agentic ecosystem produces new frameworks, protocols, and products weekly, and most confusion about them comes from comparing things that live in different buckets. Name the bucket first, and the landscape gets simple.

Bucket one: agent orchestration

This is everything about how an agent thinks and acts: model choice, tool calling, planning, permissioning, memory, and the harnesses that coordinate multi-step work across sub-agents. Frameworks like LangGraph, CrewAI, and the Agent Development Kit live here, and MCP, which connects agents to tools and data, is this bucket’s signature protocol.

It is where most of the industry’s attention lives, and where most of the well-known names compete. When people say “the agent stack,” they usually mean this bucket alone.

Bucket two: connecting agents to user-facing applications

The second bucket looks like last-mile plumbing and turns out to be one of the hardest problems in the stack. The reason: agents break the request-response paradigm the internet has been built on since its inception.

This is the layer the AG-UI protocol standardizes, and where CopilotKit builds: open-source infrastructure for connecting any agent to any user-facing application, with pre-built components or fully headless UI. The core of the bucket is the infrastructure of connectivity, not UI components.

Bucket three: agents that learn in production

Shipping an agent is the start of its lifecycle, not the end. The third bucket is self-improvement: once an agent is in production, it should get better from every interaction rather than plateau at launch quality.

The learning signals are the interactions themselves: agent-user conversations on one side, in-app user activity on the other. Learning is scoped per user, per group, and per organization, and it lands in two forms: skills injected back into the agent through in-context learning, with no fine-tuning pipeline required, and datasets exported for fine-tuning when you want them. Every learned behavior stays auditable, tracing back to its source. This is the bucket CopilotKit Intelligence occupies.

Why the buckets matter

The buckets compose; they do not compete. Your orchestration framework, your connectivity layer, and your learning loop are orthogonal choices, which is why protocols like MCP (bucket one) and AG-UI (bucket two) work together rather than fight.

The model also dissolves most category confusion. “LangGraph vs. CopilotKit” is not a comparison; it is one tool from bucket one and one from bucket two, typically used together. When a new agentic product launches, the first useful question is not “is it good?” but “which bucket is it in?”

FAQs

Which bucket does MCP belong to?
The first. MCP connects agents to tools and data, which is part of how an agent acts. AG-UI is its counterpart in the second bucket (agents to users), and A2A connects agents to each other, which also sits in the first bucket's coordination story.
Which bucket is AG-UI in?
The second. AG-UI is the open protocol for agent-user connectivity: it carries streaming output, mid-run steering, shared state, human-in-the-loop interactions, and generative UI between any agent and any user-facing application.
Do I need all three buckets?
In sequence. You need orchestration for an agent to exist at all, connectivity the moment real users face it, and learning once it is in production and you want it to compound instead of plateau.
Where do evals and observability fit?
Alongside the third bucket. Evals and observability measure agent behavior; the learning bucket is what closes the loop by turning those signals, plus live user interactions, into an agent that actually improves.

Related concepts

Talk to an engineer about agentic applications

Bring your use case. A CopilotKit engineer will walk you through how this applies to what you are building.