Xi’an, situated in Shaanxi province, is one of the most celebrated cities in all of China. Internationally renowned as the home of the Terracotta Army, the city served as China’s capital for over 1,000 years under 13 different dynasties. Today, it is one of China’s largest domestic tourism destinations, attracting around 330 million visitors annually.
With the rapid rise of AI, the way in which tourists interact with the city is changing. As generative AI becomes increasingly commonplace, tourists are beginning to expect conversational, personalised digital experiences when visiting cultural sites. The problem, however, is that these models do not have access to specialised historical data and so cannot deliver truly unique – and historically accurate – experiences for consumers.
This is why Huawei and Shaanxi Culture Industry Investment Group (SCG) have co-developed BoGuan, the world’s first commercial multimodal large language model (LLM) dedicated to cultural tourism. The partners spent two years compiling a 1.2 Petabyte dataset consisting of 31 million images, 4.4 million minutes of video, and 960 million pieces of structured text to form a specialised foundation for the platform. BoGuan then leverages this data, acting as a unified gateway aggregating 10 open and closed-source models (including Huawei’s PanGu).
“General large models lack specialized knowledge in museology, archaeology and history, making it difficult to meet the nuanced demands of niche scenarios,” noted Edric Chu, General Manager of Huawei’s Shaanxi Rep Office. “We completed 12 high-quality cultural tourism datasets […] using Huawei Cloud Data Engineering to process and label text, images, audio and video […] Supported by three major standards of data collection, management, and circulation, the platform transforms raw data resources into data assets with verifiable ownership.”
New cultural experiences and revenue opportunities
BoGuan underpins a wide range of cultural tourism services across the province, including AI travel assistants, multilingual tour guidance, museum interpretation, AI-generated marketing content, digital preservation of cultural heritage, and short-form video production.
In the B2B sector, BoGuan powers an AI “Video Factory” that fully automates short drama and advertisement production. Meanwhile, for consumers, the model powers ‘Xiaoqi’, an AI travel companion on the GO-SHAANXI app, and the Zhiying Camera mini program, which allows tourists to instantly merge their photos with AI-generated historical settings.
Since its pilot launch in September last year, Huawei says applications powered by BoGuan have reached more than four million users, and it is already delivering a major revenue boost for SCG.
“With the same team size, thanks to these technologies[…] from last year to this June our revenue has increased by roughly 30–40%,” explained Jin Yan, Chairman of the Digital and Intelligent Culture Technology Group at SCG. “The sales of related digital collectibles and creative products have exceeded 2 million Chinese yuan ($300,000).”
SCG has already begun collaborating with tourism authorities in Xinjiang and Guizhou to replicate the BoGuan framework, with the goal of potentially exporting it nationwide in future.
Network infrastructure foundation
Of course, to successfully deploy BoGuan at scale requires powerful network and data infrastructure. Thousands of simultaneous AI interactions require both massive data throughput and dense computing capabilities.
To solve this, SCG’s runs on an on-premise 48P computing platform built with Huawei’s SuperPoD architecture. It uses Huawei’s UnifiedBus technology, an interconnect protocol for SuperPoD that interconnects physical servers so that they can learn, think, and reason like a single logical server.
At the same time, China Telecom Shaanxi and Huawei have blanketed Xi’an’s Grand Tang Mall with a 5G Advanced network capable of delivering downlink speeds of 3.5 Gbps and uplink speeds of 600 Mbps. This, the partners explain, is crucial to support the scale of AI use cases being accessed throughout the region.
“The Grand Tang Mall is a textbook example of a high-traffic, high-concurrency, and high-interaction scenario, which poses an immense challenge for any network,” explained Wang Hao, Director of Mobile Communications Network Business at China Telecom Shaanxi. “Within this area […] we have deployed 46 base stations within this compact area, a density that far exceeds standard regions.”
These base stations are also equipped with AI-powered intelligent control boards that can trigger service acceleration protocols when AI photo generation is taking place.
“The embedded AI intelligently and dynamically allocates network resources to match your real-time demands,” Wang said, noting that the network can support 23,000 concurrent users during holiday traffic surges.
Bridging the AI talent gap
Beyond the technology itself, another key them to emerge from discussions about BoGuan – and about creating purpose-built LLMs for other vertical industries – is the lack of workers skilled in both AI usage and their specialist field.
“The industry faces a deficit of millions of interdisciplinary professionals who bridge the gap between cultural tourism and AI,” said Chu. “Shaanxi alone faces a talent gap of 30,000 to 50,000 in culture tourism plus AI, and the national gap is expected to exceed 1 million by 2030.”
As a result, further developing BoGuan and similar LLM projects faces a significant talent bottleneck that can only be overcome through largescale training projects.
“SCG, Huawei, and over 10 institutions have jointly established a talent training base […] aligning vocational certificates with Huawei AI certifications and university micro-majors,” explained Chu. “We expect to train over 1,000 people this year and more than 3,000 next year.”
Proof of vertical AI success
The success of BoGuan is a demonstration of how purpose-built AI models can create new revenue opportunities for telecom operators by combining connectivity, cloud, AI compute and industry expertise into a single commercial platform. Rather than competing solely on network capacity, operators must move to monetise vertical AI through managed services, industry-specific applications, and data products. With its unified AI infrastructure, platform-level scalability, and training on vertical-industry data, the BoGuan LLM helps industries tackle bottlenecks in content production, cost management, capability integration, and data security—ensuring that AI can be practically deployed to fuel sustainable business growth. This same model, in turn, serves as a blueprint for telecom operators seeking to generate lasting growth in the AI era.” Artificial intelligence is not simply a stack of technologies,” concluded Chu. “[In Shaanxi] it has become a key enabler that can activate thousands of years of cultural heritage, reshape travel experiences, and inject new momentum into the industry.”
Also in the news
TELUS and L-SPARK give Canadian startups access to AI supercomputer
Belden to acquire RUCKUS Networks for $1.85bn
VMO2 taps Suffolk solar farm for 10 years of clean energy
此内容由惯性聚合(RSS阅读器)自动聚合整理，仅供阅读参考。 原文来自 — 版权归原作者所有。