On-site photoWhile AI has demonstrated remarkable capabilities in automation, data analysis, and creative tasks, the true test of its potential lies not in what it can replicate—but in what it can meaningfully enhance, said industry experts at the NEX-T Summit 2025 co-hosted by NextFin.AI, GALA (Global Asian Leaders Alliance), Shanda Group, Barron's China and TMTPost.At the "Next-T Demo Show: Ignite the Next-Where visionary startups spark the future" roundtable, investors and founders engaged in a spirited debate over AI's ability to replace human labor in practical settings. While automation and intelligent agents are advancing rapidly, participants agreed that human judgment, empathy, and accountability remain critical.Bei Zhang, VP of Growth at Tanka, explained that while AI has become a capable assistant in handling complex tasks, it still requires human oversight in many domains. "We've trained AI to be an intelligent assistant that can complete part of a task, but many functions still need the guidance of human experts," Zhang said.Tanka is developing a vertical paid platform to manage such hybrid workflows. For instance, AI can help users write investor decks or presentations, but actual fundraising still depends on authentic communication with real investors. "We're building tools that connect users to genuine experts—AI helps, but it doesn't close deals," Zhang added.In the financial world, William J. Wu, CEO and Co-founder of Menos AI, takes a cautious stance. "AI is a machine—it doesn't bear responsibility," he said. Menos AI operates as a decision-support system that screens thousands of stocks according to user-defined logic, leaving the final call to professional analysts. "AI can help improve efficiency and scale, but users still need to be skilled investors," Wu emphasized.Arvin Sun, Founder and CEO of Traini, offered an example from the pet industry. For dog owners, he said, real trainers are often more effective than AI alone. "At Traini, we offer both—human trainers for hands-on interaction and AI models for understanding canine behavior and communication," Sun said. "When trainers can't respond in real time, AI fills the gap."Beyond replacing human roles, another major discussion centered on whether AI agents could replace SaaS platforms—the backbone of enterprise software for decades.Hunter Guo, Co-founder of Solvea, described the shift in simple terms: "Traditional SaaS is a tool, but AI agents are skilled assistants with domain knowledge." Solvea has developed an AI agent designed for customer service, powered by a deep knowledge base that continuously learns and optimizes workflow performance.According to Guo, this evolution redefines business models: "Our pricing is outcome-based—we charge based on the results we deliver, not subscription usage. This makes AI agents a combination of product and service."However, in heavily regulated and fragmented industries such as finance, challenges persist. Each financial institution operates its own set of SaaS "islands"—from equity lending to private credit—creating inefficiencies. Wu of Menos AI noted that "AI agents are effectively replacing the humans who search for opportunities across these silos."He pointed to a simple insight: "Whenever software has an ‘export data' button, that's where business opportunity lies." When a SaaS tool can't fully handle a task and users resort to Excel, it signals a gap AI can fill. "AI agents should integrate with SaaS, not destroy it. Together, they can evolve into something entirely new," Wu said.Still, commercialization challenges remain. Convincing clients to grasp the difference between SaaS and AI agents is no small task."AI agents will replace SaaS. SaaS is a business model; agents are a paradigm shift. AI delivers outcomes, not just interfaces." In Sun’s view, the future lies in paying for results rather than access—a change that could redefine enterprise software economics.In healthcare, such transformation will take time. Zifan Jiang, Chief Medical Officer at Theta Health, noted that many physicians still rely on paper records, and even electronic health systems remain fragmented. "Neither SaaS nor AI agents can revolutionize healthcare overnight," Jiang said. AI applications in medicine must achieve near-perfect accuracy in diagnostics, and building the infrastructure to support that level of reliability presents a unique opportunity of its own.While AI tools such as ChatGPT have already reshaped everyday life and work, industry leaders at the summit argued that the AI era is still in its infancy.Lenjoy Lin, Co-founder of Genspark.ai, said his company releases new products or major updates every few weeks to keep pace with AI's explosive progress. Yet, he believes society is "still at the beginning" of this journey." In the future, people shouldn't have to spend time learning how to use tools. Tools should help them focus on strategy," Lin said. Genspark.ai aims to democratize access to professional AI tools globally, especially in regions still underserved by advanced technology.Nan Liu, CEO of Quantea, reflected on how far the industry has come since the release of ChatGPT three years ago. "From CPUs to GPUs, from storage to networks, everything has changed," he said. "AI is accelerating the internet's evolution—like moving from dial-up to fiber optics."For Wendy Zhou, CEO of ADAS ECO, the contrast between 2022 and 2025 is striking. "Back then, when I asked ChatGPT how to safely transport 200 boxes, it gave me a basic checklist," she recalled. "Now, AI checks inventory, prints labels, and delivers detailed guidance—it's a complete solution. The progress is astonishing."Despite the promise, founders agreed that the path ahead for AI startups remains steep."Think big, start small," said Sun. "Begin with a narrow, deep vertical—one meter wide, one hundred meters deep. Build around small, high-quality datasets, then expand." Success, he added, comes from focusing deeply before scaling broadly.Wu of Menos AI advised startups not to "fight the dragons," referring to AI giants like OpenAI or Google. "Instead, become their allies," he said, citing DocuSign as an example. Although Adobe could have easily replicated DocuSign's core functions, DocuSign succeeded by mastering its vertical and earning user trust—eventually partnering with Adobe rather than competing against it. "No one owns the entire market," Wu said. "Startups win by connecting the small dots and growing stronger from there."As AI continues to evolve, its impact on industries, workflows, and human roles is inevitable. Yet, the question isn't whether AI can replace humans or SaaS—but how it can redefine the relationship between intelligence, responsibility, and value creation.From augmenting human expertise to reshaping enterprise models, AI is pushing boundaries that once seemed fixed. But as nearly every speaker at the summit agreed, this transformation has only just begun."We're still at the dawn of AI," Lin said. "The next era won't be about humans versus machines—it will be about how the two evolve together."更多精彩内容，关注钛媒体微信号（ID：taimeiti），或者下载钛媒体App