As we step into the year 2025, the world of artificial intelligence is witnessing a remarkable transformation, shifting from the initial stage of hardware advancements to a new era of comprehensive application development, reminiscent of the evolution we saw from the iPhone launch to the thriving App Store ecosystemThe rise of domestic large models, particularly DeepSeek, signals a game-changing moment in the AI landscape akin to the legendary fire stolen from the gods by Prometheus, igniting a new wave of innovation.

Goldman Sachs, a prominent Wall Street investment bank, views this spark as a pivotal moment where AI development transitions from building robust hardware infrastructures to focusing on application layers that enhance operational efficiencyWithin this context, AI agents have emerged as the most promising frontier for DeepSeek's innovationsThe DeepSeek-R1 model exemplifies a powerful reasoning system adept at intelligent analysis and inference, significantly shortening the path towards creating intelligent systems for enterprises seeking to maximize their operational capabilities.

The technology sector has already earmarked 2025 as the "Year of Intelligent Agents." Just before DeepSeek's rise to prominence, OpenAI's CEO Sam Altman shared in his end-of-year reflection that by 2025 we might witness the first AI agents entering the workforce, fundamentally transforming business productivityThis prediction has further elevated expectations within the tech industry for intelligent agents, which many believe will become central to the next great leap in artificial intelligence.

With these trends unfolding, businesses face pressing questions about how to shape their own strategies and make crucial decisions that will unlock the full potential of intelligent agentsAs the landscape of AI applications evolves, corporate clients are increasingly incentivized to invest in large models, driven by the need to achieve cost savings and enhanced profit margins

Advertisements

Intelligent agents present a compelling opportunity to reduce labor costs while simultaneously leveraging AI to boost productivity.

Take, for instance, McKinsey's utilization of automated process allocation within intelligent agents, allowing them to compress project processing times from 20 days down to just twoSimilarly, Momentic has replaced traditional quality assurance (QA) teams with QA testing agents, while JD Cloud’s JoyCoder can predict code, generate annotations, and conduct intelligent code reviews, boosting productivity for thousands of developers at JD by 30%. According to Relevance AI, each set of intelligent agents can handle workloads equivalent to that of five full-time employees, highlighting the immense commercial potential of AI agents.

The transformative impact of intelligent agents has led Microsoft’s CEO Satya Nadella to boldly predict that AI agents might eventually supplant Software as a Service (SaaS). While both operate through software to automate specific tasks, intelligent agents embody a new layer of productivity, merging AI capabilities with SaaS tools to directly address enterprise challengesFor example, a Customer Relationship Management (CRM) system could integrate a sales agent that automatically generates sales reports or forecasts customer churn risks, demonstrating the potential for deep integration across various business functions.

The efficiency gains offered by intelligent agents suggest that reshaping SaaS through AI applications is not just beneficial but imperativeThe dynamics of this shift are particularly palpable in China, where the integration of large models could pave a healthier path for the SaaS industry, tackling issues like excessive software customization through enhanced generalizability and replicate capabilities of AI agents.

From **NoNoise**, insight into a diverse array of B2B enterprises reveals a growing consensus around developing intelligent agents, though strategic clarity can be lacking

Advertisements

An inherent tension exists between IT and business units, with numerous internal business scenarios and a plethora of SaaS tools, making it unclear where to begin innovationDeloitte’s "2025 Technology Trends" report advises companies to focus on identifying one or two high-value AI application scenarios.

JD Cloud has echoed this sentiment, asserting that delving deeply into a few key areas is more critical in this phase of development than spreading resources too thinlyThe shift from a multitude of scattered agents toward concentrating efforts on the most promising core scenarios is necessary for maximizing returnsThe rise of potent reasoning models like DeepSeek and ChatGPT-O1 enhances the ability of intelligent agent products to align with these core scenarios, allowing businesses to explore market demands more extensively.

In the drive toward innovation, some strategies emerge as crucialJD Cloud emphasizes that leveraging the "80/20 rule"—where 20% of core applications satisfy 80% of general requirements—must inform the development of intelligent agentsFive core scenarios identified include collaborative office tools, specialized assistants, customer service, marketing, and data applications—each aligning with fundamental business needs to optimize the utility and value of AI.

Successful SaaS companies like Salesforce and Zoom prove this concept with their dominating market shares, achieved through deep service integration and product utilityFor instance, Salesforce revolutionized the CRM landscape by pioneering cloud integration, while Zoom addressed interoperability challenges in video conferencing, leading to significant adoption and loyalty in usage.

JD Cloud's internal practices validate these principles, boasting the development of over 10,000 intelligent agents spanning multiple sectors—including retail, logistics, finance, health, and computing—resulting in substantial enhancements in workforce efficiency

Advertisements

The marketing sector exemplifies this with JD’s ad placement AI Agent effectively acting in various roles to assist merchants, allowing for streamlined management across multiple accounts with mere verbal commands.

During significant sales events like the 618 shopping festival, data showed companies like HP could reduce manpower by over 20% for promotional tasks owing to these intelligent assistantsIn international trade, intelligent agents like JD’s CMS HScode generator enable vast efficiency gains, slashing processing times for customs codes dramatically, from extensive manual searches down to single-step workflows activated by simple triggers.

In the realm of collaborative tools, JD’s JoyLaw quickly assesses and mitigates contract risks, condensing review times of lengthy documents to mere seconds and boosting legal team productivity significantlyLikewise, customer service agents powered by intelligent systems can address user queries around the clock while successfully resolving a vast majority of issues without human intervention.

JD Cloud’s focus on these practical implementations reflects a systematic effort to build robust intelligent agent capabilities within its platforms, ensuring readiness to meet core sector needs swiftlyThe resulting language-based intelligent agent platform serves as a comprehensive solution that enterprises can configure readily without the need for extensive trial and error.

Yet, as organizations gear up to deploy these intelligent systems, it’s essential to recognize that the path to effective AI integration will vary across enterprise typesLarger organizations may benefit from creating a unified intelligent agent platform that encapsulates all AI tools within modifiable modules, enabling agile growth and iteration based on evolving business needsConversely, smaller enterprises might opt for direct access to public cloud services, capitalizing on the existing frameworks established by leading AI service providers.

With major cloud service providers rolling out their own intelligent agent platforms, organizations can align their strategies with options suited to their unique operational contexts

Advertisements

Advertisements