3 UX considerations for a multi-agent system

 

This chapter covers

  • Examining how the user interface landscape has evolved from the command line to multi-modal, multi-agent systems
  • Tracing the shift in software engineering practices from traditional development to autonomous multi-agent systems
  • Discerning between end user and developer personas in understanding the user
  • Comparing interactive versus offline multi-agent scenarios
  • Spelling out UX design principles for multi-agent applications, including capability discovery, cost-aware action delegation, interruptibility, observability and provenance.

The ultimate goal of building a multi-agent system (or any software tool) is to solve a specific problem for a user. This system may be utilized as components by developers (such as software developers via SDKs or APIs etc.) to create larger systems or directly used by end-users to accomplish particular tasks. Regardless of the target user, it is crucial that we develop systems with an optimal user experience - systems that are effective, meaning they help users achieve their tasks with minimal friction or effort, and enjoyable, making them engaging and delightful to use. To achieve this, we must understand the user and their behaviors, motivations, attitudes, and goals, and then design systems that cater to these needs.

3.1 Why UX Matters

3.2 User Interfaces - From Command-Line to Multimodal, Multi-Agent Interfaces

3.2.1 Command-Line Interface (CLI)

3.2.2 Graphical User Interface (GUI)

3.2.3 Immersive Interfaces (AR/VR)

3.2.4 Intelligent Multimodal (Natural Language, Images, Video, Audio), Multi-Agent Interfaces

3.3 From Software 1.0 (rules) to Software 3.0+ (Autonomous Multi-Agent Systems)

3.3.1 Software 1.0: Traditional Software Development

3.3.2 Software 2.0: Machine Learning and AI

3.3.3 Software 3.0: Generative Models and Autonomous Multi-Agent Systems

3.4 The AI Reliability Paradox: Generative AI Models Being Capable but Not Always Reliable

3.5 Understanding the User: End User, and Developer Personas

3.5.1 End Users

3.5.2 Developers

3.6 Multi-Agent UX Design Principles

3.6.1 Capability Discovery

3.6.2 Cost-Aware Action Delegation

3.6.3 Observability and Provenance

3.6.4 Interruptibility

3.7 Interactive vs Offline Multi-Agent Scenarios