AI Empowers Government Digitalization: Smart Office Deployment and Its Value

As China pushes forward with its digital transformation agenda, local government offices — the front line of public administration — are grappling with heavy document workloads, high data security demands, and complex cross-agency coordination. Carriers, together with BotSmart, have developed a full-stack, homegrown smart office solution powered by Huawei's FTTO intelligent computing appliance, which has already been deployed in local government settings. By deeply integrating software, hardware, and carrier ecosystems, the solution not only resolves the core challenges facing local public sector operations, but also demonstrates a clear path to commercializing AI + B2B offerings, creating a replicable model for digital and intelligent transformation across industries.

Scenario-driven: Strategic alignment between local government needs and carrier transformation

Efficient operations in local government departments are crucial for enhancing governance and optimizing public services. However, technological and resource constraints create three key pain points. First, processing a large volume of documents is time-consuming. Second, government information is sensitive and therefore requires stringent protection. Third, interdepartmental collaboration needs improvement. Meanwhile, carriers are pushing for B2B service transformation by leveraging cloud-network synergy to meet government and enterprise clients' needs with a comprehensive solution that integrates connectivity, computing power, security, and AI. In this context, carriers, BotSmart, and Huawei have formed a deep partnership.

BotSmart is an AI company that specializes in vertical applications of large models. The company offers AI content production platforms built around content security. Together with Xinhua News Agency's State Key Laboratory of Media Convergence Production Technology and Systems, BotSmart developed a full-stack AI platform, Xinhua Miaobi, for official document writing (as shown in Figure 1). Xinhua Miaobi integrates large models with authoritative government data corpora to provide five core capabilities: searching, writing, proofreading, question and answer (Q&A), and learning. The platform now supports over 200 types of official documents and 94 governance scenarios. With features such as AI-powered writing and intelligent proofreading, it improves document quality and efficiency while ensuring authority and compliance and precisely meeting the needs of local government offices.

Figure 1: The application interface of Xinhua Miaobi

Huawei's FTTO intelligent computing appliance is a software-hardware integrated device that hosts AI capabilities. As the industry's first all-in-one solution combining all-optical networking, intelligent computing, local storage, and security, it delivers intelligent networking, computing, application, and maintenance.

By pre-installing Xinhua Miaobi's AI capabilities on the FTTO intelligent computing appliance, carriers can offer a robust, plug-and-play, full-stack smart office solution to drive digital transformation in local government offices. This precise alignment of technology with vertical scenario needs serves as a critical enabler for carriers to implement their AI + B2B strategy.

Journey breakdown: From needs insight to scaled replication

To commercially deploy AI + B2B solutions, carriers follow a four-step process: pinpoint pain points, tailor solutions, validate through pilots, and scale for replication. The collaboration among one carrier, BotSmart, and Huawei exemplifies this path. Staying closely aligned with customer needs, this approach ensures practical, replicable AI solutions.

Deeply understanding needs and pinpointing pain points

A community in Zhongshan, Guangdong province, serves as an example. The carrier conducted field research in this community and surrounding local government offices, identifying core office pain points and real-world needs. This led to three core objectives: Improve governance efficiency, ensure secure and controllable operations, and enable seamless collaboration. Integrating the workflows of local government staff, public service demands, and government compliance rules, the carrier developed a solution that features AI writing, intelligent proofreading, authoritative databases, and collaborative office tools. To address security requirements for government data, the carrier set full-stack localization and on-premises data control as core requirements, combining the strengths of Huawei's FTTO intelligent computing appliance with Xinhua Miaobi's localized secure storage capabilities.

Co-building the ecosystem and tailoring the solution

Leveraging its ecosystem integration capabilities, the carrier developed a full-stack, homegrown smart office solution that combines cloud-network infrastructure, AI capabilities, and scenario-specific applications. Powered by Xinhua Miaobi, a content security large model, Huawei's FTTO intelligent computing appliance, and cloud-network resources, the solution weaves together technology, resources, and use cases. Its architecture has three layers.

• Cloud-network infrastructure: Using gigabit government private lines and elastic cloud resource pools, the FTTO intelligent computing appliance delivers all-optical networking with low latency and high reliability, ensuring secure data transmission and scheduling of computing power. Equipped with Huawei's high-specification computing cards, the appliance offers 100–560 tera operations per second (TOPS) of AI computing power and supports local deployment of 7B to 32B foundation models.
• AI capability integration: The solution merges Xinhua Miaobi's core AI capabilities with the intelligent inference capabilities of the content security large model to enhance Xinhua Miaobi's government document processing performance. Key features like AI writing and intelligent proofreading help tackle common challenges of local government document handling. The solution refines writing styles based on local needs to eliminate clichés and empty phrases, while its long-text generation capabilities can produce documents exceeding 10,000 words.
• Security and operations and maintenance (O&M) assurance: The solution provides on-premises data management with 99.99% reliability, building a strong security defense for government data. The FTTO intelligent computing appliance supports remote management via information technology managed services (ITMS), allowing automatic fault reporting and rectification, which significantly simplifies O&M and reduces costs.

During the critical phase of joint solution customization, BotSmart and Huawei tackled common industry challenges, including resource constraints on edge devices and difficulties in local deployment of large models. To ensure seamless integration between the FTTO intelligent computing appliance and cloud computing resources, BotSmart and Huawei conducted specialized adaptation and performance optimization for the 14B and 32B device-side content security models. Employing advanced techniques like model quantization and compression, operator optimization, memory scheduling and reconstruction, and lightweighting of the inference engine, the teams significantly cut compute and memory usage while maintaining accuracy in core scenarios like AI writing and intelligent proofreading. Multi-round concurrent pressure testing and E2E latency optimization ensured stable performance under high loads. Government document generation and proofreading latencies remained within acceptable ranges, enabling efficient and reliable model operation on the FTTO appliance.

This collaboration not only strengthened the full software-to-hardware chain, but also enabled deep integration of technical solutions, engineering delivery, and service processes, paving the way for scaled replication and multi-scenario expansion.

Validating through a commercial pilot and refining the solution

The carrier chose a community in Zhongshan as the pilot site, deploying the solution with partners for a three-month trial. This community represents the typical challenges of local government services, combining the complexities of urban-rural governance with the essential needs of public service delivery. It is China's first full-stack homegrown smart office system built on the FTTO intelligent computing appliance. The project demonstrated the solution's core value, provided quantifiable and replicable experience, and laid a strong foundation for scaled implementation.

During system operations, BotSmart and Huawei gathered customer feedback to refine the solution and upgrade applications, targeting user pain points to improve the solution's adaptability and usability.

• Optimizing writing style for service scenarios: To address community feedback that AI-generated official documents were too formal and disconnected from real-world needs, the teams optimized the model's writing style and introduced a lighter, more conversational, community-friendly model. The upgraded model is suitable for non-official documents, such as work reports, resident applications, and social media posts, ensuring both compliance and practicality.
• Simplifying the operational process: To address feedback that the generation process was too complex, the teams streamlined the entry for scenario-specific writing. Users can now enter one-sentence prompts to generate documents. Optional fields can be skipped with one click, lowering the barrier for users and meeting the need for lightweight, fast writing.
• Jointly optimizing model inference efficiency: To address the issue of taking three to five minutes to generate 400–500 words, BotSmart and Huawei deeply optimized the inference engine for the 14B and 32B models, fine-tuned computing power scheduling, and accelerated transmission links on the FTTO hardware platform. This reduced the time to generate 500 words to under one minute, solving the efficiency problem at the edge.
• Upgrading local knowledge base capabilities: To address inaccurate answers from the knowledge base, the teams optimized the retrieval-augmented generation (RAG) algorithm, semantic matching rules, and Q&A policies for vertical scenarios. This improved the accuracy of answers to frequent questions about community policies, service procedures, and resident issues, while supporting lightweight use of non-confidential documents.

Standardizing and replicating to expand value

Following the successful pilot, the carrier streamlined the project experience and developed a standard implementation model combining carrier deployment, ecosystem partner enablement, and government use. This model was then rolled out to local government departments nationwide. The carrier also leveraged its nationwide service network to offer end-to-end services, covering deployment, training, and O&M. This has significantly lowered the barrier to digital transformation, enabling rapid and effective deployment of the solution.

Value creation: From operational efficiency to ecosystem win-win

This solution demonstrates the value of the carrier's AI + B2B approach, delivering tangible and measurable benefits to the public, government departments, and carriers alike — realizing the principle of "technology creates value, and value fuels growth".

For citizens: Simplifying services with one-tap, one-stop processing

Serving the public is the core goal of local government digital transformation. The solution streamlines processes and improves service efficiency, making public services more accessible. Xinhua Miaobi's efficient document processing ensures swift communication of government directives, enabling timely policy implementation and meaningfully improving the public service experience.

For government: Boosting quality, efficiency, and security while building expertise

The solution fundamentally changes the traditional mode of handling government documents at the local level. The AI writing function cuts document writing time by more than 70%. The intelligent proofreading function lowers the error rate to less than 0.1%. Together, these boost administrative efficiency and governance while meeting security and compliance requirements.

The solution also enables one-click document generation, proofreading, and retrieval, freeing staff from heavy document processing and policy search. Its built-in government policy knowledge base and case library provide real-time policy interpretation and service guidance, helping staff quickly build their expertise.

For carriers: Strengthening B2B transformation and completing the ecosystem

This collaboration has brought stable government and enterprise customer relationships for the carrier, demonstrating the core competitiveness of their AI + B2B approach and helping the carrier transition from a connectivity provider to a strategic partner of digital and intelligent transformation. By integrating BotSmart's software with Huawei's hardware, the carrier has further developed their AI + B2B ecosystem while setting a benchmark for scaling the solution across local government offices. This aligns with the carrier's strategy of "going deep on industry use cases to build high-quality, scenario-specific models", giving them a competitive edge in the global B2B market.

In summary, the joint effort between the carrier, BotSmart, and Huawei clearly demonstrates the commercial logic of AI + B2B solutions in the government and enterprise market: customer needs at the core, cloud-network synergy as the foundation, AI capabilities as the engine, and ecosystem collaboration as the support. Through a process of needs insight, solution development, pilot validation, and scaled replication, the solution delivers both technology empowerment and value creation.

As AI technologies and cloud-network infrastructure continue to converge, the B2B space is poised for new growth. As an ecosystem partner, BotSmart will continue to strengthen AI capabilities, helping carriers leverage their position as core builders of digital infrastructure. The success in local government offices will be replicated across more vertical industries, building a new ecosystem for high-quality digital economy development.