2025-07-18T20:40:06z
- Openai COO Brad LightCap defined AI agents in a recent episode of its podcast.
- He said agents need to “autonomously” perform tasks they have never seen before.
- This week, Openai announced its latest agent. This agent worked with ChatGpt to perform the task.
There is not much agreement among AI leaders about how to define all the new things they are developing.
For example, no one really agrees about what artificial general information is. Microsoft CEO Satya Nadella recently called it “non-sensical benchmark hacking.”
There is also debate about exactly what an AI agent is.
Brad Lightcap, Chief Operating Officer of Openai, provided his thoughts on Openai's podcast this week.
“It must be a system that can take on autonomously and ensure that you can hand over complex tasks that can be done with a high level of proficiency that you've never seen the job before,” he said. “I've never seen it before.” part teeth “Critical,” added Lightcap.
On Thursday, Openai announced its latest contribution to the space: ChatGpt Agent. The new agent “can now work with its own computer to handle complex tasks from start to finish,” Openai said in the announcement. This combines the work of operators to engage with the website and integrate information, as well as ChatGpt's conversational abilities, the company said.
Companies are doing their best to integrate agents into their work. Large tech companies from Intuit to Salesforce are rolling out them at the speed of light. The startup has also announced new agents.
The race to adopt AI made workers worried about whether they would soon replace machines. However, at least in Lightcap's view, humans are necessary at all levels.
Agents aren't just entities “trained to copy,” he said. They utilize their reasoning ability to solve problems.
Still, it's just a “teammate,” he said.
