OpenAI CEO Sam Altman showed a new AI agent on Tuesday.

He called it "remarkably capable, always-on."

This happened one day after the company stopped the rollout of a stronger model.

They stopped it because of safety worries.

At OpenAI's yearly developer conference, Altman announced several products and updates.

One was new agents called Dots.

These agents can do ongoing tasks for users on their own.

The agents compete with Meta's personal agent Muse.

Muse became very popular after Meta's conference last week.

Altman said Dot agents would be "like an AI helper that always has your back."

On Monday, OpenAI said it was delaying the other model.

Its researchers had raised concerns.

Many people in the industry also want AI development to slow down.

They want safety measures to catch up.

Altman did not talk about these concerns in his main speech.

But in a question-and-answer session, he said the company is spending more on safety, security and watching AI agents.

He said the AI boom is more like a renaissance than an industrial revolution.

OpenAI also announced GPT-6.1 Sol.

It is an upgraded version of its GPT-6 Sol model.

OpenAI also announced a faster, more expensive tier called "Ultrafast."

These announcements came just before leaders of six big AI companies signed a voluntary agreement with President Donald Trump on Tuesday.

The agreement says companies should "self-police" their development.

It calls for "robust internal controls," an "independent external auditor" and a board committee at each company.

Altman did not go.

But OpenAI president and co-founder Greg Brockman was at the White House.