DEEPZERO (02723.HK): Named a Leader in IDC's Worldwide B2C Enterprise Agentic Marketing 2026 Vendor Assessment, the Only Chinese Company on the List

NewTimeSpace News: In September 2026, International Data Corporation (IDC) published the IDC MarketScape: Worldwide B2C Enterprise Agentic Marketing 2026 Vendor Assessment, in which DEEPZERO (stock code: 02723.HK) was placed in the Leaders category and became the only Chinese company among the vendors listed in the report. Over the past two years, the most watched question about enterprise AI was whether it could generate content; now that large models have become a baseline capability, competition at the application layer is shifting to a harder question: whether AI can enter a company's real business, take part in judgement, execution and review, and ultimately affect operating results. That is where the value of DEEPZERO's inclusion in IDC's Leaders category for global B2C enterprise agentic marketing lies.

The subject of the IDC assessment is not all marketing software that uses large models, but platforms able to create, deploy and manage marketing agents and let them execute tasks autonomously or semi-autonomously in real business environments. A product in scope must connect enterprise data, marketing channels and business systems, support multi-step, multi-agent workflows, and provide production deployment, governance, audit, human-machine collaboration and human override mechanisms; Copilots that only generate content, deliver data insights or complete a single task do not fall within IDC's definition of an agentic marketing platform. This change in criteria marks the central dividing line in the commercialisation of enterprise AI. If AI only produces a piece of copy, it solves a content production problem; if it can identify audiences from customer data, judge the right moment to reach them, design marketing paths, call channel systems to complete delivery and then adjust the next step of strategy based on conversion results, it is taking part in enterprise decision-making. The former lifts the efficiency of one link, while the latter changes an entire business chain. Companies therefore no longer ask only how smart a model is, but further ask whether it can run reliably, whether problems can be traced, and whether the results it brings can be measured.

What sets DEEPZERO apart is that it did not begin looking for AI application scenarios only after the large-model boom. The company was founded in 2009 and its long-term accumulation centres on complex decisions in marketing, sales and customer operations; over the past decade and more it built a full data and algorithm foundation around customer data, marketing automation, advertising delivery and performance optimisation, and the emergence of Deep Agent reorganised those existing capabilities into an agent system for enterprise decision-making. The IDC report states that Deep Agent was launched in February 2025, covering six decision domains (product innovation, market entry, social marketing, advertising, sales and customer service, and CRM), with more than 30 specialised agents, using a dual-layer architecture that combines large language models with proprietary domain machine learning technology. AI VOC, AI Consumer, AI Survey, AI Persona, AI Ad Campaign Manager and AI Sales Assistant correspond respectively to consumer feedback analysis, user insight, market research, user profiling, advertising delivery and sales support, connected behind the scenes to enterprise data, business rules, industry knowledge and executable processes. In the case of an automobile brand, the store visit rate of high-propensity users rose to about 2.7 times, which shows that the value of an agent system lies not only in producing analysis reports but in linking user identification, delivery actions and business results.

Huang Xiaonan said in an interview that what companies really need to solve is an operating question such as which products should be pushed to which of one million members and at what time, and she compared large models to a newly graduated PhD: broad in knowledge, but not necessarily understanding a particular company's customers, processes and historical data. This explains why DEEPZERO insists on combining large models with domain models, with large models providing general understanding, reasoning and interaction while domain models take on prediction, identification and optimisation. In marketing decisions, repurchase probability, conversion probability, churn risk and budget performance cannot be achieved by general language capability alone; they also require industry data, enterprise data and business feedback accumulated over a long period.

The value DEEPZERO has accumulated over the past 17 years is not only the volume of data but the decision experience built around it: how to define a customer, how to identify effective signals, how to break business goals into model goals, how to turn model results into channel actions, and how to keep correcting the system with operating feedback. Its business structure has formed a two-way loop: one end is agent products for internal enterprise processes that help clients build their own AI decision-making capability, and the other is agent-centred marketing services that directly take part in client operations and deliver business results; problems arising in service scenarios in turn drive platform and model iteration, while improvements in platform capability provide the basis for scaling the services. In IDC's global assessment, DEEPZERO appeared in the same market research alongside international vendors including Adobe, Salesforce, Microsoft, Oracle and SAP. For a company from China with the application layer at its core, entering the same list means its products are beginning to be compared against global enterprise software standards: whether they can connect complex systems, handle real data, establish governance mechanisms, keep running inside a client organisation, and ultimately prove that investment can be converted into operating results. IDC also pointed out, however, that agentic marketing is still at an early stage and that enterprise buyers are still working out its budget ownership, ROI measurement and organisational impact. For a company based in China and developing for the global market, how to expand global market coverage, keep proving commercial returns and build long-term trust in accuracy, compliance and brand safety remain questions for the next stage.