Ticket triage · Production

Building Featest: extending AI agent interfaces with A2UI and the AVC Pattern

The problem

AI agents capable of reasoning and planning are forced to communicate through simple chatbot interfaces and basic markdown, limiting their ability to express rich, interactive user experiences.

First attempt

A pure AI-First approach proved too slow for basic, repetitive tasks because generating UI tokens on the fly adds latency that a hand-optimized native app does not.

Workflow diagram · grounded in source
1
User interacts with Lit client
trigger
“User interacts with the Lit Client, which uses the official @a2ui/lit renderer. It acts as a state machine, processing SurfaceUpdate events to patch the DOM efficiently.”
2
Client sends intent via A2A
integration
“Client sends intent via A2A to the Backend. The A2UIClient wraps the user's input (text or events) in a standard JSON-RPC envelope.”
3
Controller Agent processes logic
ai_action
“This agent handles the business logic. It doesn't care about pixels. It inputs a user request, decides which tool to use (e.g., vote_feature, add_comment), and outputs structured data.”
4
View Agent formats A2UI JSON
ai_action
“This agent is the designer. It takes the data from the Controller and translates it into A2UI JSON. It cares about layout, typography, and hierarchy.”
5
Client renders streamed UI
output
“Backend Agent processes logic and streams back A2UI JSON instructions. Client renders the UI components dynamically.”
Reported outcome

A hybrid model mixing static UIs for core workflows with dynamic agentic components for complex, intent-driven tasks is recommended; A2UI's progressive streaming makes the application feel incredibly responsive.

Reported metrics
Application responsivenessincredibly responsive
Reported stack
A2UIA2AGoogle Agent Development Kit (ADK)@a2ui/litSequentialAgentLitFlutter
Source
https://mlops.community/blog/building-with-a2ui-extending-the-expressiveness-of-ai-agent-interfaces
Read source ↗

Frequently asked questions

What did this team achieve with this AI workflow?

A hybrid model mixing static UIs for core workflows with dynamic agentic components for complex, intent-driven tasks is recommended; A2UI's progressive streaming makes the application feel incredibly responsive.

What tools did this team use?

A2UI, A2A, Google Agent Development Kit (ADK), @a2ui/lit, SequentialAgent, Lit, Flutter.

What results were reported?

Application responsiveness: incredibly responsive (source-reported, not independently verified).

What failed first in this deployment?

A pure AI-First approach proved too slow for basic, repetitive tasks because generating UI tokens on the fly adds latency that a hand-optimized native app does not.

How is this ticket triage AI workflow structured?

User interacts with Lit client → Client sends intent via A2A → Controller Agent processes logic → View Agent formats A2UI JSON → Client renders streamed UI.

Related ticket triage cases