ZTE Company, a worldwide supplier of built-in info and communication know-how options, has introduced that its chief improvement officer, Cui Li, spoke at AI Innovation Asia 2025, hosted by Economist Influence, the thought-leadership arm of The Economist Group.
On the panel “How Could AI Assist You? Agentic AI and the Buyer Expertise”, Cui Li shared ZTE’s imaginative and prescient for agentic AI and illustrated how it’s already reshaping buyer expertise and operational fashions throughout the corporate. She additionally underscored its function in strengthening resilience, enhancing oversight and accountability, and urged organisations to arrange now for the period of agentic AI.
Beneath is Cui Li’s Q&A Part
Q1: Setting the scene – How is agentic AI impacting buyer expertise in your sector?
Agentic AI is redefining the consumer expertise—not solely about UI design, however from response to understanding and co-creation. For ZTE, below our “AI for All” technique, we’re integrating agentic AI into 4 key areas: networks, computing, houses and private units.
For instance, we allow degree 4 and past autonomous networks with three engines—Nebula Telecom Giant Mannequin, large information and digital twin. In actual follow, ZTE and China Cellular co-created multi-agents that may detect community issues and allow self-healing, slicing troubleshooting time by 47%.
Q2: Strategic transformation – How will growing digital autonomy via agentic AI reshape how organisations construct resilience and adaptableness?
Truly, we’re in a extremely unsure period now. This requires us to begin with the top in thoughts, specifically, at all times search stability from uncertainty and construct a robust moatlike the snowball impact. Additionally we should keep agile to identify modifications and pivot on a dime, evolving from a machine-like organisation into an natural, adaptive one. After which we have to understand how AI can actually assist us. Giant AI fashions are already acting at or past PhD degree. Whereas brokers take a step additional by integrating reminiscence and instruments, performing because the real-world utility of fashions. And agentic AI can coordinate totally different brokers to automate extra complicated, time-consuming duties. In fact, this is a perfect standing. The reality is, whether or not brokers or agentic AI, they’re nonetheless at a really early stage, and face many technical challenges. However given the hypergrowth of AI, I imagine that options might be proper in place very quickly.
After which I imagine AI can exert results solely after we keep it up to the lengthy haul. Intelligence is constructed on digital and community foundations. With out digital transformation, an organization is not going to actually obtain intelligence, to not point out turning into extra resilient or adaptable—it’s identical to attempting to run earlier than you’ll be able to stroll. Additionally, going clever requires data engineering, course of restructuring and an AI mindset, which is a marathon, not a dash.
At ZTE, our digital journey began in 2016, after which clever transformation in 2022. Right here is our expertise: infrastructure first, however maintain a stability between {hardware} and software program; perform systematic planning from the highest down, to make sure everyone seems to be on the identical web page; make steady funding, as disruptive breakthroughs come from each small step; and at last, begin from high-value and concrete eventualities, then iterate quick to handle any uncertainties alongside the best way.
Q3: Oversight and accountability – As AI techniques make extra choices autonomously, how can firms keep oversight, guarantee accountability, and protect digital sovereignty?
In a nutshell, to maintain people within the loop. Duties like design, assessment, decision-making and supervision—they nonetheless should be performed by individuals, who ought to stay in the end accountable. Automation is only a means, not an finish. What people ought to actually fear about will not be being changed—it’s stepping again or being absent on this course of.
Each coin has two sides, so do these fashions. Their generalisation, emergent skills and steady evolution—sure, these are the options of a really game-changing tech. However they inherently include hallucinations, black-box issues and so forth. As well as, human has social intelligence and morality, one thing that AI can by no means actually grasp. So basically, AI is constructed on statistical fashions—it lacks frequent sense in the actual world, not to mention navigating complicated trade-offs like people do.
Extra importantly, rolling out AI in a enterprise wants the deep integration with know-how. We’ve to think about components like accuracy, safety, compliance and division of accountability, and take note of each workflows and KPIs. From ZTE’s hands-on expertise, I’d like to present some suggestions: First, firms ought to develop their very own data engineering tasks and domain-specific massive fashions—plus RAG and digital twins to be skilled and dependable. Second, establish the concrete issues that brokers want to unravel. A one-size-fits-all agent usually finally ends up doing nothing properly. Third, know when to make use of brokers or workflows. Brokers are higher at dealing with complicated duties with variable execution paths, whereas workflows are extra correct and environment friendly in extremely predictable eventualities. Final, allow end-edge-cloud collaboration to make sure each value effectivity and safety. Whereas amongst all these key components, people are nonetheless those steering AI in the proper path and creating actual worth.
This fall: Trying forward – How do you see agentic AI evolving from task-based automation to built-in enterprise companions, and what instant actions ought to organisations take to future-proof themselves for agentic AI?
From a tech perspective, we will consider brokers or agentic AI as proactive digital staff. Past easy or repetitive duties, they’ll join whole workflows, realise cognitive automation, and even self-evolve. Brokers now work properly in eventualities which can be well-structured, info-heavy, fault-tolerant, and have clear suggestions loops. However they usually get caught within the lab when the real-world environments develop into extra complicated or dangerous. So like I stated, brokers and agentic AI are nonetheless early. Within the subsequent 12 months or two, they’ll primarily deal with vertical industries. After that, they’ll tackle complicated duties with higher autonomy, getting extra generalised, adaptive and in a position to study and evolve. Brokers now are creating very quick. Gemini 3, which was simply launched final month, units a brand new bar for AI fashions with SOTA reasoning, multimodal understanding, and agentic capabilities.
For organisations, I feel embracing AI is the one technique to go. Deploying AI isn’t nearly connecting to APIs—it’s reshaping processes, constructions and groups. Firms have to first make mid- and long-term plans, and be adaptable sufficient to maintain up with tech and market modifications. Subsequent, begin from high-value, business-specific eventualities, after which iterate quick. That’s how we will actually grasp AI. Additionally, it’s reshaping our expertise technique. Sooner or later, three forms of expertise will matter most: AI specialists, who drive this know-how ahead; AI energy customers, who foster innovation and improve effectivity; and individuals who transcend AI with high-order pondering and a wholesome mindset. Lastly, to present AI full play, firms ought to restructure themselves and plan for a way forward for “human-AI symbiosis”.
AI Innovation Asia 2025 is a high-level dialogue platform connecting enterprise leaders, know-how pioneers and policymakers. Centred on 15 in-depth thematic classes and insights from greater than 40 business specialists, the platform focuses on the commercialisation pathways of frontier applied sciences, equivalent to generative AI and Agentic AI, to assist companies flip technical perception into tangible development and navigate sustainable digital transformation throughout Asia-Pacific’s complicated market panorama.
Touch upon this text through X: @IoTNow_ and go to our homepage IoT Now
