AI agents, advanced software programs that act autonomously to perform specific tasks without human assistance, have become the latest buzzword in tech and business.
Nvidia’s Vice President of AI Software, Kari Briski, told Entrepreneur that AI agents will change how we think about our daily tasks, from writing code and research papers to managing calendars and payroll.
“I think agents will transform the way we work,” Briski said.
Kari Briski.
According to a Franklin Templeton 2025 technology outlook paper, AI agents have a variety of practical use cases for everyday life, from acting as tutors to planning meals to booking travel. In business, they can automatically ask and respond to customer questions by looking up answers in internal documents.
Meanwhile, companies from Salesforce to Nvidia have introduced their own AI agents with the intention of helping workers and customers make their lives easier.
But how does the technology actually help businesses?
Related: Nvidia CEO Jensen Huang Says the New ChatGPT Needs ‘100 Times More’ of His Company’s AI Chips
Briski said that entrepreneurs and founders who want to start their own companies could tap into AI agents to write code and build business plans. Agents can even act as a “sparring partner” to brainstorm ideas or think through a situation.
“If you’ve ever noticed that you’re a little bit more productive when you can spar off somebody, agents can provide that to help you be more productive,” Briski said. “Instead of having to go seek out an expert who might not have time, [it’s] able to iterate now.”
Briski also noted that AI writing code has come a long way. In the past, the tech was focused on code completion, where AI finishes a line of code as programmers type it out. Now, AI agents write full programs on their own, making software developers “10 times more productive,” she says. This frees up time for the developer to focus on their system architecture, product, and end-user experience.
The technology can also help employees research markets faster by surfing the web, finding sources, and writing thorough research papers based on a prompt — complete with citations to the web sources they consulted to craft the paper.
Competitor OpenAI, for example, demonstrated its ChatGPT deep research agent earlier this year, which can create detailed reports with nearly 30 sources in about 10 minutes. OpenAI relies on Nvidia chips to power its AI products, though it is working on developing its own AI chips by the end of the year.
Briski noted that over the past two years, Nvidia’s technology has saved ServiceNow, a cloud-based platform that helps businesses create and automate workflows, over 400,000 hours of human labor.
“AI agents offer complex decision-making capabilities and the ability to interact with and influence the environment,” Briski said.
How much do AI agents cost?
Nvidia did not disclose how much its software for enterprises to make AI agents costs, but IT support company Transputec estimates that the technology can cost anywhere from $0.99 per conversation for simpler tasks to $30 per hour for more complex ones.
Salesforce AI agents are priced at $2 per conversation and can be used for tasks like customer service and employee support. OpenAI is reportedly preparing PhD-level AI agents priced between $2,000 and $20,000 monthly. The company’s ChatGPT Pro paid plan with deep research capabilities costs $200 per month.
“AI agents that are available as a service start off very expensive because they’ve got very big infrastructures and services,” Briski said.
Still, the high price tag hasn’t stopped companies from using AI agents. Accenture, Datastax, and Oracle all use Nvidia’s AI to build agents for their companies.
