If 2025 marked the year generative artificial intelligence (AI) entered the mainstream, then the hottest buzzword in the AI field for 2026 is undoubtedly "AI Agent." The industry widely believes that this year will witness the rapid adoption of AI agents, which are poised to evolve from mere assistive tools to collaborative partners capable of independent action, reasoning, and executing tasks.
With advancements in technology, AI has been steadily transitioning from simple "chatbots" to "AI agents" equipped with autonomous planning, decision-making, and execution capabilities. On Dec. 31, 2025, US tech giant Meta announced its acquisition of Manus, a Chinese AI agent startup, preemptively launching the opening salvo of the 2026 AI agent race. While the specific terms of the deal have not been disclosed, reports suggest the acquisition price may have reached as high as US$2 billion, making it Meta's third-largest acquisition ever, trailing only WhatsApp and Scale AI. Meta now joins the ranks of Silicon Valley giants such as Google and Microsoft in investing heavily in AI agents.
Emergence of AI Agent companies
In fact, as early as last year's developer conference, Google announced plans to fully integrate AI agents into Google Search and its AI assistant, Gemini. Microsoft has similarly accelerated its development in this domain, embedding a new AI agent in the taskbar of Windows 11, with plans to transform Windows into a core platform for AI agents.
According to a report by data analytics firm CB Insights, among the 1,500+ technology markets it tracks, five of the top ten most active markets in 2025 were related to AI agents. One in five new unicorn companies is focused on AI agent technology. Additionally, half of the top 20 AI agent startups by revenue were founded less than three years ago.
From a business perspective, 25% of AI agent startups established since 2023 have achieved commercial distribution levels that traditionally took over five years for startups to reach. AI agents have especially demonstrated significant value in enterprise applications. Among 64 surveyed organizations, two-thirds stated they plan to adopt AI agents for customer support within the next 12 months.
Consulting firm Gartner notes that 2025 marked a pivotal year as AI agents evolved from simple assistive tools to complex collaborative ecosystems. With foundational AI model providers increasingly launching platforms or products incorporating agent functionalities, AI agents are fundamentally changing how enterprises automate complex tasks and decision-making processes.
Breaking free from "chatbot" mode
As AI agents continue to mature, Silicon Valley's top venture capital firm, a16z, predicts that the "natives" of the internet are shifting from humans to AI agents. A report by CB Insights suggests that while AI agents in 2025 mostly operated under "constrained" conditions—limited to structured workflows and predefined frameworks—2026 will see AI agents advancing toward "full autonomy." These agents will be capable of completing complex decisions, self-adapting to changes, and executing tasks without human intervention. Moreover, their forms will transcend existing chatbot interfaces, expanding to web browsing interfaces, voice-driven applications, and other diverse formats.
In addition, multi-agent systems are expected to become increasingly mainstream. For example, the multi-agent orchestration platform CrewAI, backed by venture capital firm Insight Partners, has already been adopted by 40% of Fortune 500 companies. On the commercial front, Gartner estimates that while only about 1% of enterprise software featured built-in AI agent functionality in 2024, this proportion could soar to 33% by 2028. By then, approximately 15% of daily business decisions are expected to be automated by AI.
As we step into 2026, the "collaborative" capabilities of AI agents will become ever more prominent. AI agents are no longer tools or machines designed to replace humans but collaborative partners for individuals and enterprises. This "human-machine collaboration" is expected to generate substantial wealth. Goldman Sachs forecasts that by 2030, AI agents could create approximately US$7 trillion (around HK$54 trillion) in economic value.
(Source: Ta Kung Pao)
Related News:
Meta accused of 'copying' Chinese AI models
Tiiny AI reveals world's smallest personal AI supercomputer, verified by Guinness World Records
Comment