331
Productivity & Workflow350
Automation & Workflow232
Software Development240
Marketing & Growth193
Writing & Content Creation196
AI Infrastructure & MLOps182
Photography & Imaging153
Design & Creative168
Data & Analytics155
Customer Support145
Education & Learning133
Sales & Outreach128
Voice & Speech135
Operations & Admin107
Sierra CEO Bret Taylor says people may soon use AI agents to get tasks done by writing requests in plain language, instead of clicking through apps.
In short: Sierra CEO Bret Taylor says many workplace apps may fade into the background as people use AI agents through plain language instead of clicking menus.
Bret Taylor, co-founder and CEO of Sierra, told attendees at the HumanX conference in San Francisco that the way people use software is likely to change soon. Taylor previously served as co-CEO of Salesforce.
He argued that many business tools are used only occasionally, which makes them annoying to learn. He gave the example of Workday, which many employees open only during onboarding or for benefits enrollment. Instead of hunting through screens and buttons, Taylor says people will increasingly just type what they want in normal sentences.
Sierra recently launched a product called Ghostwriter. Sierra describes it as an “agent designed to build other agents.” An AI agent is software that can take actions for you, like a helpful assistant that can do steps in other systems. Sierra’s idea is that a person describes a task, and Ghostwriter creates and deploys a specialized agent to do it.
Taylor said Sierra used Ghostwriter to build an agent for retailer Nordstrom in four weeks. Sierra also reported strong business growth in earlier updates, including reaching a $100 million annual revenue run rate less than 21 months after it was founded.
Some investors and technologists told TechCrunch that AI agents are still not fully hands-off today. Many companies, including Sierra and legal AI startup Harvey, use “forward-deployed” engineers, meaning staff who work closely with customers to keep these agents updated and working correctly. The key question is how quickly these agents can become reliable with less ongoing human help.
Source: TechCrunch AI