In June, Shanghai’s high-temperature warnings arrived at the Expo Center simultaneously with the AI热潮 (AI craze).
Even on a Friday (6/20) afternoon, a weekday, the security lines were long—an endless stream of technology builders, corporate decision-makers, and developers filled the venue. People sat on the floor in front of the broadcast screens for the Golden Hall keynote speeches, long queues formed at the VR glasses in the exhibition area for experiences, and appointments with tech giants were booked solid in seconds. All these phenomena undeniably proclaimed one fact: the “Year of Implementation” for generative AI is sweeping across industries with concrete scenarios.
If you’re also curious about how generative AI is moving from the lab into industry, and how Chinese tech companies are leveraging the cloud to go global, this conference report might offer some insights.
01 Technical Mainstays: The “Iron Triangle” of AGENT, Multimodal, and MCP
Flipping through the summit handbook, three keywords frequently popped up:
- Agentic AI: From sub-forum naming to 60% of technical topics, AI agents took center stage. IKLOD’s digital intelligence practices in vertical industries, Goldwind Science & Technology’s large meteorological forecasting model, and even an AI judge assistant in the legal field – when AI upgrades from executing commands to autonomous decision-making, the productivity revolution truly begins.
- Multimodal Breakthroughs: Amazon Nova Sonic’s anthropomorphic voice engine caused a sensation. Hualai Technology restructured its cost base using multimodal models, while Agora achieved an interactive revolution where “voice is the interface.”
- MCP Architecture: As Amazon Web Services’ secret weapon, MCP permeates the entire development toolchain, migration optimization, and security protection scenarios – MCP has become the central nervous system for generative AI implementation.
02 Industry Implementation: From Proof-of-Concept to Ton-Level Capacity
What impressed me most at this conference was the deep application of GenAI across various industries. When Anker Innovations CIO Gong Yin demonstrated how AI optimizes global supply chain decisions, when Fosun Pharma used AI Agents to shorten new drug R&D cycles, when TCL workshop fault response time was compressed from hours to minutes – AI is no longer suspended in PPTs but is deeply embedded in the industrial capillaries.
Below are some application examples from the exhibition area for those who need them:
- Generating Reports with Multiple Agents: A multi-Agent workflow system can automatically process content extraction and integration from various file formats (PPT, PDF, Word, and images), and automatically write reports according to the end-user’s required format and specifications. The workflow includes but is not limited to: Document Summary Agent, Optical Character Recognition (OCR) Agent, JSON Format Conversion Agent, and Report Filling Agent. This is suitable for scenarios such as technical document generation, seamless integration of customer requirements with proprietary template data, and automatic generation of reports compliant with bidding template specifications.
- Contract Extraction and Review: Manufacturing contracts often span hundreds of pages and contain hundreds of manufacturing parameters, making contract review time-consuming and difficult. The contract analysis and review system utilizes a customized Agent Flow, where each page of the contract calls multiple large models for precise extraction and summarization, facilitating use by contract reviewers and designers, ultimately achieving over 95% recall and accuracy.
- Intelligent Bidding and Parts Matching Assistant: In supply chain scenarios, enterprises face challenges such as low efficiency in processing purchase orders, complex parts matching, and cumbersome bidding processes. Amazon Web Services has built a generative AI-powered bidding assistant solution asset, based on Amazon Bedrock and Amazon OpenSearch. This system supports intelligent generation and analysis of tender documents, automatically processes purchase orders, utilizes large language models to understand technical specifications and procurement requirements, and accurately matches suitable parts and suppliers from a vast catalog. The solution significantly improves procurement efficiency, reduces costs and error rates, shortens the demand-to-delivery cycle, and continuously optimizes enterprise procurement strategies and supplier management through knowledge accumulation.
- Intelligent Medical Content Generation Center: Medical Insights Hub (MIH), an intelligent medical content generation center, uses generative AI to provide intelligent medical writing, querying, verification, and translation services. It achieves professional document management through various large models and agents supported by Amazon Web Services, and supports API extensions to other applications (such as Microsoft Word, etc.) to meet usage needs. Relying on a cloud-native architecture, it assists in enterprise internal and external content management, accelerates drug launch, and promotes medical innovation. The platform connects to knowledge bases such as PubMed, covering the entire process from clinical research to commercial transformation, integrating medical terminology libraries and customized templates to meet high-quality translation, query, and document verification needs, helping unleash the value of medical data and maintain industry competitiveness.
- GraphRAG-based Financial Information Analysis Engine: In current financial information analysis processes, investors face the challenge of quickly obtaining effective information from massive amounts of data and making decisions. In traditional methods, investors manually process and analyze complex information such as announcements, investment relationship changes, and financial reports, which is inefficient, prone to errors, and easily leads to missed information and judgment biases. The GraphRAG-based intelligent financial information analysis engine innovatively integrates Retrieval-Augmented Generation (RAG) with knowledge graphs, achieving intelligent correlation and semantic understanding of multi-dimensional information. Built on services such as Amazon Bedrock, Amazon Neptune, and Amazon OpenSearch, it achieves intelligent information processing through generative AI technology and provides dimensional correlation analysis through knowledge graphs, significantly improving the efficiency and accuracy of investment decisions.
Contract Metric Extraction
03 Going Global: China AI’s Springboard for Globalization
“Global Expansion” became the concluding topic for several sub-forums:
- Deye Technology showcased its green energy global solution, using AI to solve EU carbon tariff compliance challenges.
- Huolala leveraged generative AI to optimize its overseas logistics marketing chain, reducing customer acquisition costs by 37%.
- The gaming forum was even more of a focal point: Roblox built a multi-model metaverse platform based on Amazon Web Services, and Chinese short-form dramas are breaking through cultural barriers for overseas expansion with the help of AI translation.
When Sinnet (West Cloud Data) declared “reshaping the digital future with generative AI,” technology’s global outreach has ascended to a contention for rule-making power.
04 Ecosystem Carnival: The Triumph of the Developer Economy
Another interesting group at this conference were the “geeks” and “vibe coders,” as they might call themselves. But I think they are courageous pioneers and doers, because no matter when, action always trumps words.
- Developer Lounge: Hands-on challenge attracted hundreds for real-time coding; the Amazon Q CLI practical zone had winding queues.
- Certified Elites Meetup: The “Golden Jacket” battle unfolded – genius developers with all 12 AWS certifications were chased for photos.
- 1000 Aldea Application Program: 30 teams developed intelligent applications using Amazon Q Developer, ranging from legal document generation to quantitative trading systems, validating the explosion of low-code AI development paradigms.
05 Final Thoughts: From Technical Architecture to Productivity Revolution
As I was leaving, I overheard a manufacturing CIO remark, “Last year we were still discussing large model parameters; today, we’re calculating ROI.”
This perhaps reveals the deeper value of the summit: when AI moves from the lab into the workshop, when developer toolchains connect with enterprise pain points, technology finally sheds its mythical halo and evolves into a measurable, reusable new form of productivity.
As Andy Jassy, CEO of Amazon, once said: “We are on the cusp of an Agentic AI explosion.” Increasingly, more people are flocking to this AI grand event, taking away not only technical insights but also action blueprints for transforming code into business value and turning concepts into global competitiveness. The tech feast will eventually end, but this efficiency revolution and overseas expedition, ignited by AI, is just beginning.
Tips for Attending Exhibitions:
Just like traveling, it’s best to plan ahead when attending exhibitions and forums. You can find official website information or check promotional channels like Xiaohongshu to get sufficient information (including parking and surrounding amenities). This way, you can make the most of your time and obtain the most valuable content in a limited period.
I am Yutou Xiaobao. Follow me to continue exploring the growth universe of GenAI.
If you found this article helpful, don’t forget to like, bookmark, and share it!










