By: Devika Bhalla
The user interface as we know it is on the brink of transformation. In today’s digital ecosystem, humans are no longer the dominant audience online. A study by DesignRush estimates that nearly 80 percent of all web traffic now comes from bots rather than people — meaning much of the content and interfaces designed for “users” are increasingly being consumed, parsed, and reshaped by machines.
This shift is rapidly extending into the enterprise world. Salesforce notes that “AI agents are poised to transform user experience design from creating interfaces for human users to orchestrating interactions between humans and agents.” In other words, the primary users of enterprise systems are evolving from employees to AI agents that execute tasks, exchange information, and coordinate processes.
READ: Metaverse: Sony embraces virtual world, others are not far behind (May 19, 2022)
Dharmesh Shah, CTO of HubSpot, summed it up succinctly: “Agents are the new apps.” According to IDC’s Future Enterprise Resiliency & Spending (FERS) Survey from February 2025, more than 80 percent of enterprises believe AI agents are replacing traditional packaged applications as the new system of work.
The implications are clear. UI and UX can no longer be designed solely for humans clicking buttons and filling forms. Instead, they must evolve into systems that allow humans to oversee, arbitrate, and trust the autonomous agents doing the work.
When traditional UI falls short
Consider the largest expense management systems that are used in large enterprises currently. Today, the process is still entirely human-centric. Employees manually upload receipts from Uber and hotels, enter project codes, reconcile transactions, and submit reports for approval. Managers then review these line by line. It’s rigid, form-driven, and places the burden on humans to stitch together context across multiple systems.
Now imagine the same process in an agentic system. The AI agent automatically pulls data from Uber, hotels, and email, reconciles it with corporate card feeds, applies company policy, flags exceptions, and prepares a draft report for a manager to review. In this model, the human’s role shifts from manual entry to supervision — highlighting why traditional interfaces can no longer keep pace.
Rigid workflows become inefficient in an agentic environment, where flexibility and traceable decision paths are essential. Trust takes precedence over speed, especially in areas like finance, where managers must understand an agent’s reasoning and verify data provenance. Workflows are no longer linear, as agents span multiple platforms and systems. And while chat-based UIs may offer convenience, wrapping a legacy app with a chatbot interface doesn’t solve deeper issues of orchestration, context, and knowledge integration. As Infosys argues, true agent process automation requires intelligence layers — intent, context, orchestration, and knowledge.
Redefining design principles
Salesforce and Infosys outline several emerging principles that define what a truly agentic interface should be. Future systems will be built around intent-first design — focusing on what users want to accomplish rather than prescribing every step. They will support cross-platform orchestration, allowing agents to collaborate across applications, APIs, and services.
Real-time capability discovery will become crucial, enabling interfaces to adapt dynamically based on available agents and services. Transparency will also be central: humans need to know which agents are active, what they are doing, and when intervention is required. Infosys further stresses that agentic automation succeeds only when supported by multiple layers of intelligence — intent, context, orchestration, and knowledge — all working together to ensure control and trust.
What the future of UI looks like
In the agentic era, interfaces will be built on agent-native foundations, designed with the assumption that the primary user is an AI agent. Design will move away from linear user journeys toward intent mapping and orchestration across systems.
Human governance will remain critical. People must remain the final authority — able to pause, redirect, override, or approve an agent’s actions without disrupting the broader workflow. Clear signals and audit trails will ensure compliance and accountability.
Explainability and trust will define success. Every agent action should be traceable and understandable in plain language, with full transparency into data sources, reasoning, and alternatives considered. Role-based visibility will help operators, managers, and regulators access the right level of insight.
Interoperability will also be key. As multiple agent systems emerge, standardized UI protocols will be required to allow agents to pass context, data, and intent reliably between platforms. Governance and safety frameworks will ensure that these interactions remain secure and consistent.
READ: Innovating video conferencing: Meet Hour’s CTO says it has edge over industry giants (June 27, 2024)
Finally, future UIs must be adaptive and multimodal. Interfaces will shift dynamically based on user role, context, and device — spanning screens, voice interfaces, mobile components, and immersive environments like AR and VR. The best designs will balance human-friendly clarity with machine-readable semantics.
The next frontier
Enterprise interfaces were once built to guide humans through every step of a process. In an agentic world, they must be re-engineered to allow AI agents to work autonomously while giving humans the tools to monitor, audit, and intervene when necessary.
The winners of this transformation won’t be the companies that design the sleekest dashboards, but those that create systems where agents can work effectively — and humans can govern confidently.
(Devika Bhalla works as a product strategist at a Fortune 500 tech company and is a Distinguished Member of the American Society of AI. The opinions expressed in this article are her own.)