Sierra CEO Bret Taylor says the age of button-based interfaces is ending

Bret Taylor, co-founder and CEO of Sierra, believes that the way people interact with software is on the verge of a major shift, moving away from traditional button-driven interfaces toward natural language-based systems.

The company recently introduced Ghostwriter, an AI agent designed to create other agents. Through this “agent as a service” approach, Sierra aims to replace conventional click-based applications with systems where users describe what they need. Ghostwriter can then generate and deploy a specialised AI agent to handle the task autonomously.

Taylor argues that this shift is particularly relevant in enterprise environments, where many software tools are used infrequently. Speaking at the HumanX conference in San Francisco, he pointed out that employees often log into platforms like Workday only during onboarding or for occasional tasks such as benefits enrollment. Instead of learning complex interfaces, he suggests users will increasingly rely on natural language to complete tasks without directly navigating the software.

“I truly think that’s where the world is going,” Taylor said, emphasising his belief that conversational interfaces will replace traditional user interfaces in many scenarios.

He also noted that Sierra is already using Ghostwriter internally to deploy AI agents at high speed. For example, Taylor said the company implemented an AI agent for Nordstrom in just four weeks.

Sierra has seen rapid growth since its founding. The company announced last fall that it had reached a $100 million annual revenue run rate in under 21 months. It was most recently valued at $10 billion following a $350 million funding round led by Greenoaks Capital in September.

Taylor emphasised that businesses are less interested in building software itself and more focused on solving problems. “Most companies don’t want to make software,” he said. “They want solutions to their problems.”

Despite this vision, not everyone in the industry believes the transition will happen quickly. Some technologists and investors note that current AI agent systems still require significant human involvement. Companies offering these solutions, including Sierra and legal AI startup Harvey, often rely on engineers who continuously monitor, update, and fine-tune the agents to ensure they perform as expected.

While fully autonomous AI-driven software may still be some distance away, Taylor remains confident that natural language interfaces will play a central role in the next generation of computing.