The AI industry is continuously developing more intelligent models; however, the real challenge lies in engineering solutions that make these models “cost-effective”—ensuring simple requests bypass large models, avoiding redundant computational expenses, and optimizing every unit of computing power for maximum efficiency.

With this in mind, our focus turns to Li Yanming, Senior Enterprise Solution Expert at Volcano Engine, who has spent years at the forefront of AI implementation within leading enterprises. His expertise centers not on “what AI can achieve” but on transforming AI from a cost center into a driver of genuine productivity.

Conference Overview

We will explore these questions in-depth at the offline conference on October 23.

Hosted by OFweek Vicor Network · Artificial Intelligence, the “WAIE 2026 (10th) AI Industry Conference & AI Agent Industry Application Conference” will be held on October 23, 2026, at The Westin Shenzhen Nanshan.

This conference will focus on AI agents, covering both foundational technological innovations and real-world industry applications. It will offer detailed insights into national AI industry policies, general agent architectures, and challenges in commercialization. The event will dissect technical iteration hurdles and paths for cross-industry integration in high-demand areas, while sharing benchmark cases across sectors. Hundreds of government and enterprise leaders, research experts, R&D vendors, and key decision-makers from manufacturing, energy, government services, healthcare, education, finance, commerce, and more will gather. Deep discussions will revolve around technological innovation, industrial chain collaboration, and industry-academia-research transformation, aiming to connect government, enterprise, industry, academic, and research resources, deliver actionable agent solutions, showcase innovative products and technical solutions, help enterprises enhance brand influence, expand collaboration opportunities, and fully accelerate the large-scale implementation of AI agents across industries.

Keynote Speakers and Topics

Recently, Li Yanming, Senior Enterprise Solution Expert at Volcano Engine, confirmed his participation in the conference, where he will deliver a keynote speech titled: “When AI Moves from ‘Novelty’ to ‘Normalization,’ the Real Efficiency Revolution for Enterprises Has Just Begun.”

In his speech, Li will shift the focus away from “how powerful large models are” and instead analyze real enterprise cost structures post-AI adoption: Which expenses rise after full AI integration? How should models be tiered and selected? How can traffic be intelligently routed, and where should caching be applied? Additionally, he will explore how product stacks—from Doubao Work to Trae, and from Coze to HiAgent—can transition AI from “pilot projects” into normalized, scalable productivity.

Keynote Speech: When AI Moves from ‘Novelty’ to ‘Normalization,’ the Real Efficiency Revolution for Enterprises Has Just Begun

Volcano Engine, ByteDance’s cloud and AI service platform, provides products and services that directly leverage ByteDance’s expertise in growth, technology, and operations. Its offerings include: Doubao Large Model for multimodal capabilities, Doubao Work for AI-enhanced daily office tasks, Trae for R&D coding, and Coze with HiAgent for building and deploying agents into business processes. Through serving tens of thousands of enterprises, Volcano Engine has developed not just model access methods but a comprehensive approach to computing power selection, traffic routing, and cost management—enabling AI to transition from “demonstrable” to “daily usable” within enterprises.

Speech Highlights

01　Why Costs Spiral After Full AI Adoption
Why does the cost curve steepen when scaling from a “100-person pilot” to “full-scale daily use”? The first step is identifying where expenses originate before discussing savings.

02　Model Tiering and Selection: Keep Simple Requests Away from Large Models
Not all tasks require flagship models. Determine which tasks should be handled by lightweight models versus large models, and establish clear boundaries.

03　Intelligent Routing and Caching: Minimize Redundant Computations
Enterprises face daily repetition of similar requests. Identify, reuse, and eliminate redundant computational costs.

04　A Unified Product Stack: Make AI Normalized Productivity
From Doubao Work to Trae, and from Coze to HiAgent, how should products be arranged and integrated to ensure optimal computing power utilization?

Speaker Introduction

Li Yanming
Senior Enterprise Solution Expert, Volcano Engine

◆　Deeply involved in AI-driven intelligent transformation across industries, leading AI strategy implementation for top financial institutions.
◆　Extensively participated in AI tool adoption practices for employees in leading enterprises, developing systematic methodologies for transitioning large models from technical validation to scalable productivity.
◆　Specializes in agent architecture design, computing power resource planning, cost management, and deep integration of AI across office and R&D workflows.
◆　He believes the ultimate goal of AI is not to showcase technical prowess but to enable every enterprise to afford, effectively use, and derive real business value from AI.

Conference Information