WPS Software Accelerates Smart Office Innovation with Generative AI, Deploying AI Features in Just 2 Months

WPS Software, a global provider of office software with over 35 years of experience, serves millions of users worldwide through its flagship product, WPS Office. Recognizing the rapid advancements in AI technology and the potential for intelligent assistance within office software, WPS Software identified an opportunity to explore generative AI applications that could position the company as a leader in smart office innovation. In 2023, the company launched its “All in AI” initiative, focused on integrating generative AI into its products to create a more intuitive and intelligent user experience.

However, implementing generative AI in office software presented several complex challenges. WPS Software needed to evaluate and select suitable language models, manage model accuracy by reducing hallucinations, and adapt prompts effectively to ensure high-quality user interactions. Additionally, ensuring compatibility across platforms, optimizing AI model costs, and maintaining global data security and compliance were critical. The company also faced a shortage of AI talent and the need for rigorous performance evaluation to uphold quality and efficiency standards in generative AI applications.

Since the start of its overseas operations, WPS Software has relied on AWS to support over 60 percent of its international infrastructure, including WPS Office, the WPS international website, and desktop versions of its applications. This solid foundation with AWS provided WPS Software with the stability needed to explore generative AI on AWS confidently.

To advance its AI-driven strategy, WPS Software introduced WPS AI for international users, integrating it directly into WPS Office as a versatile AI assistant. Powered by Amazon Bedrock, WPS AI focuses on three primary areas: AIGC for content generation, Copilot for smart assistance, and Insight for knowledge extraction. WPS AI’s core features—such as advanced spell check, content rewriting, and automated PowerPoint generation—are built using Anthropic’s Claude 3 large language model (LLM) in Amazon Bedrock. With Claude 3’s extensive context window of up to 200,000 tokens, WPS AI can deliver reliable content generation, refining, and management capabilities to users, significantly boosting productivity across everyday office tasks.

Using Claude 3 as its underlying language model, WPS AI streamlines workflows and task management, assisting users in various aspects of content creation and enhancing user efficiency. For knowledge extraction, WPS AI utilizes Claude 3 to help users analyze and interpret data, supporting informed decision-making within WPS Office. Amazon Bedrock provides WPS Software with a unified API for selecting and evaluating models, enabling WPS AI to rapidly test and deploy models suited to specific business needs. “Based on WPS Software’s business needs, we’ve found that choosing the right, cost-effective LLMs brings more benefits than pursuing the most advanced or powerful models in the industry,” explained Ning Zhang, vice president of Global Business at WPS Software. “Amazon Bedrock provides us with diverse models, making it easy to find options that align with our objectives.”

In addition to Amazon Bedrock, core AWS services such as Amazon CloudFront, Amazon Elastic Container Registry (Amazon ECR), Amazon Elastic Kubernetes Service (Amazon EKS), Amazon CloudWatch, AWS Identity and Access Management (IAM), and Amazon Simple Storage Service (Amazon S3) provide the backend infrastructure to support WPS AI’s operations. This infrastructure ensures scalability, performance, and security for WPS Software’s extensive backend needs, enabling them to deliver a robust AI experience to millions of international users.

Responsible AI and data security are also central to WPS AI’s deployment. AWS’s responsible AI practices, particularly within Amazon Bedrock, help protect the privacy of WPS Software’s international users. Data in WPS AI remains solely within WPS Office, is excluded from model training, and is encrypted in transit and at rest, ensuring that sensitive information stays secure. AWS’s compliance with more than 140 security standards and provision of over 300 security services globally further aids WPS Software in meeting regional data protection regulations.

AWS collaborated closely with WPS Software throughout the integration process, engaging science and strategy experts from the Generative AI Innovation Center and solution architects to work alongside WPS in developing a technical roadmap. This close collaboration enabled WPS Software to overcome deployment challenges and ensured the rapid, successful rollout of WPS AI, delivering a reliable, intelligent assistant that transforms productivity across office applications.