OpenAI's AI agents need to catch up

OpenAI, which popularized the modern generative AI chatbot, is falling behind in the fast-growing category of continuously running, consumer-facing AI agents. With its 2026 DevDay event approaching, the company is widely expected to release an AI agent called Aeon to compete with Meta's Muse, OpenClaw, Instinct, and Google's Gemini Spark. To succeed, Aeon would need to combine the strengths of existing platforms — such as OpenClaw's utility, Muse's affordability and ease of setup, and Instinct's conversational simplicity — while addressing security vulnerabilities that have plagued AI agents since OpenClaw's rise. AI agents capable of booking travel, buying groceries, scheduling appointments, and helping with work have been an industry goal for years, and the race has accelerated since OpenClaw went from proof of concept to practical success. Meta's Muse has topped App Store charts with 600,000 daily active users, while Instinct stands out for its ease of use across chat apps like iMessage. OpenAI employs OpenClaw creator Peter Steinberger, who is likely contributing to Aeon's development. The company is also hyping GPT-6 Astra as a frontier model that would power its agent. However, OpenAI faces integration disadvantages compared to Meta and Google's ecosystems. DevDay may also include updates on AI hardware developed with Jony Ive, GPT-6 advancements, and further AGI claims, all under pressure from the company's upcoming IPO.

(Source：The Verge)