With agency artificial intelligence, businesses do business more dynamically. Instead of traditional pre-programmed bots and static rules, agents can now “think” and invent alternative paths when unprecedented conditions arise. For example, a business domain ontology Such as FIBO (financial industry business ontology) can help to keep agents within safeguards and avoid unwanted behavior.
The bottleneck is now in the user experience (UX) layer. Although agents are dynamic and transform with ontology-guided information drift, the user interface is still very static. These practices of fixed fields and configurations can hinder the creative freedom afforded to agents. such as modern standards AG-UI (agent User Interface) It helps to facilitate communication between UX and agents, but still screens must be pre-defined at design time.
Newer technology takes this to the next level, dynamically allowing agents to display the user screen they want based on specific content. One A2UI – user interface agent. With A2UI, we first define the UX scheme for how the components will be displayed. This loosely coupled schema allows agents to build screens based on data.
Agents now communicate with an A2Ui-compatible “renderer” that dynamically renders screens based on JSON content dynamically generated by the agents. The screens are fully interactive and can interact with the respective agents using the AG-UI. companies like Copilot kit They are actively building A2UI renderers that can dynamically build the UI from a JSON specification and connect it back to the agent via AG-Ui.
Moreover, by using newer compression standards such as token object notation (TOON) can help achieve highly efficient compression and include schemas such as ontology and A2UI in context hints. Of course, as the models become more intelligent, they will also include the ability to automatically generate screens compatible with A2UI and AG-UI through pre-training.
The following diagram illustrates one view of this architecture.
As shown, the A2UI specification complements the business ontology and aims to specify the logic for user interface components. As an example of credit approval, the ontology would define business concepts such as loans, parties, interest terms, contracts or terms. This data is typically present in multiple source systems in various forms, and a common business ontology helps unify this into a common “language”." The A2UI specification will define how user experience components are to be rendered.
In the future, only the specification should be changed, not the individual screens, because the screens are created with fresh content each time. Also, since A2UI uses AG-UI under the hood, screens communicate with the original agent that created the content. So events like button clicks and form submissions can be tracked and responded to. This whole experiment takes place inside a glass bottle – for example, a traditional chatbot.
Business delivery integrates ontology, agents, A2UI JSON, dynamic content displays, and AG-UI messaging. Everything is driven by business logic and relationships defined in the ontology, meaning less is left up to interpretation by the UX designer and UI developer. We still need these roles in projects, but reusable components are only defined and built once. Rinse and repeat!
For example, you can specify that any communication message sent to the user (error, information, warning) will be displayed within a panel with your company logo and conform to ISO 9241-110. With Agent AI and A2UI, a custom agent can validate these messages and display them on the screen according to standards.
The chat interface is still your main interface for users, but the A2UI components are rendered the same way. More importantly, existing user screens can be reused as templates to dynamically create new screens. This makes your business highly resilient to business and regulatory changes.
Patterns like A2UI reduce dependency on the user interface and complement the dynamic nature of the business. Imagine a company going through an acquisition and needing to add new logos to thousands of uniforms. This logic can now be configured in the A2UI specification and ontology and UI changes will be propagated when users access the forms. It helps the business to be dynamic and increase the productivity of the employees.
Dattaraj Rao is an innovation and R&D architect at Persistent Systems.
