Inside Mirakl’s Agent Commerce Vision

Mirakl 为全球领先的零售商和品牌提供市场平台与零售媒体解决方案。随着 AI 能力的进步， Mirakl 采取了不同寻常的路径：AI 不再只是专家使用的工具——每位员工都被期望成为构建 AI 能力的一员。

我们采访了 Mirakl 的联合创始人兼联合 CEO Adrien Nussenbaum 与首席数据与 AI 官 Anne‑Claire Baschet，听他们讲述 Mirakl 如何把 AI 作为产品核心以及团队工作方式的核心。

“最初的愿景是让 100% 的 Mirakl 员工使用 AI。几个月前我们把目标调整为 100% 的 Mirakl 员工成为 agents 的建设者——为个人目的构建，或重新定义团队工作流，以为用户带来更多价值。”—— Anne‑Claire Baschet， Mirakl 首席数据与 AI 官

关键成果一览

• 使用 ChatGPT Enterprise 制作内部技术文档的速度提升约 70%
• 客户支持效率提升 37%，同时保持 96% 的客户满意度
• 依托 Mirakl 的 AI 原生 Catalog Transformer，目录上线时间缩短 91%
• 分类错误约减少 50%，提升了数据质量与转化率
• AI agents 现已实现全天候、多语言运行，满足全球支持需求

Mirakl 推广策略内部观察

Mirakl 的 AI 路线从一开始就有明确的文化导向：公司没有满足于“每个人都使用 AI”。愿景演进为“每个人都用 AI 来构建”。员工不仅消费 AI 输出——他们设计并拥有那些改变团队工作方式的 agents。正如 Baschet 所说，“我们现在要放大价值——识别跨团队或复杂的工作流，并集中领导力把这些落地。”

产品策略也反映了这一转变。 Mirakl 从把 AI 当成辅助工具，转向让 AI 在明确、结构化的工作流中代表用户行动——同时保留人为判断与细微差别的控制权。这一点在若干关键工作流中都能看到。

在内部文档上， Mirakl 的技术写作团队使用 ChatGPT Enterprise 大幅加速产品文档编写——写作时间约减少 70%，团队间一致性也得到提升。

在客户支持上， Mirakl 构建了一个以 Mirakl 文档为依据的支持 agent，效率提升 37%，同时保持 96% 的客户满意度，并能即时提供多语言支持，使员工能把注意力放在更高价值的问题上。

在目录上线上——这是最大的突破——这款 AI 原生的 Catalog Transformer 把上线时间缩短了 91%，分类错误减少约 50%，从而加快了营收实现并提升了体验质量。

来自 Mirakl 的领导力启示

• 从最难、最有价值的客户问题入手：有意义的价值会在问题大且广泛存在的地方最快显现。
• 不要等技术“成熟”再行动：在前沿构建可以比竞争者更快学习。
• 让用户掌舵：衡量成功的标准是用户是否像自己做的一样信任输出。
• AI 成功＝问题＋模型＋体验：价值来自三者协同发挥。
• 目标要超越渐进式改进：AI 不只是优化工具——设定系统级而非功能级的雄心。

“当你拥抱 AI 时，要考虑的目标——规模、时机、覆盖面和影响——应远大于你以往的规划范围。”—— Adrien Nussenbaum， Mirakl 联合创始人兼联合 CEO

下一步：迈向由 agents 驱动的商业

Mirakl 认为下一阶段的商业将由代表购物者和商家的 agents 来执行任务——涵盖发现、比价、购买、配送追踪和售后工作流。

为此， Mirakl 正在开发 Mirakl Nexus——为 agent 原生商业设计的基础设施：

• 支持多商家购物篮
• 处理复杂的交易流程
• 与零售商系统集成
• 为由 agents 驱动的购物与服务体验提供可扩展的底座

Mirakl 的角色是：提供零售商参与这波新商业浪潮所需的基础设施、连接与运营专长——而不是从头重建一切。

加入前沿

全球已有超过 1,000,000 家企业通过 OpenAI 实现了切实成果。欲了解更多，请联系团队（https://openai.com/contact-sales/）。

Mirakl powers marketplaces and retail media for leading retailers and brands globally. As AI capabilities advance, Mirakl has taken a distinct approach: AI isn’t just a tool for specialists—it’s a capability every employee is expected to build with.

We sat down with Adrien Nussenbaum, Co-founder & Co-CEO, and Anne-Claire Baschet, Chief Data & AI Officer, to hear how Mirakl is making AI core to both its products and the way its teams work.

“The initial vision was 100% of Mirakl workers use AI. We shifted a few months ago to 100% of Mirakl workers being builders of agents—for their individual purpose or to redefine workflows in their teams to bring more value to the user.”

• 70% faster creation of internal technical documentation using ChatGPT Enterprise
• 37% efficiency improvement in customer support, while maintaining 96% customer satisfaction
• 91% faster catalog onboarding with Mirakl’s AI-native Catalog Transformer
• ~50% fewer categorisation errors, improving data quality and conversion
• AI agents now operate 24/7 and in multiple languages, meeting global support expectations

Mirakl’s AI approach starts with a clear cultural direction: the company didn’t stop at “everyone uses AI.” Their vision evolved to “everyone builds with AI.” Employees aren’t just consuming outputs—they’re designing and owning the agents that change how work gets done in their teams. As Baschet puts it, “we are now about scaling value—identifying workflows that are cross-team or complex, and focusing leadership energy to make those happen.”

The product strategy mirrors this evolution. Mirakl moved from AI as assistance to AI acting on behalf of the user in clear, structured workflows—while keeping humans in control for judgment and nuance.This shows up across key workflows. 

In internal documentation, Mirakl’s technical writers used ChatGPT Enterprise to dramatically accelerate the production of product documentation—reducing writing time by around 70% and improving consistency across teams.

In customer support, Mirakl built an agent that answers support questions using Mirakl’s documentation, improving efficiency by 37% while maintaining 96% satisfaction, and providing instant, multi-language support so staff can focus on higher-value issues. 

And in catalog onboarding—the major breakthrough—the AI-native Catalog Transformer cut onboarding time by 91% and reduced categorisation errors by ~50%, enabling faster time-to-revenue and a higher-quality experience.

Start with your hardest, highest-value customer problems
Meaningful value shows up fastest where the problem is big and widely felt.

Don’t wait for the technology to “mature”
Build at the edge to learn faster than competitors.

Keep users in the driver’s seat
Measure success by whether users trust the output as if they created it themselves.

AI success = problem + model + experience
Value depends on all three working together.

Aim beyond incrementalism
AI isn’t just an optimization play—set ambition at system-level, not feature-level.

“When you embrace AI, think about objectives—scale, timing, reach, impact—that are far greater than what you’re used to planning for.”

Mirakl believes the next phase of commerce will involve agents acting on behalf of shoppers and merchants—spanning discovery, comparison, purchasing, delivery tracking, and after-sales workflows.

To enable this, Mirakl is developing Mirakl Nexus—infrastructure designed for agent-native commerce:

• Supports multi-merchant baskets
• Handles complex transaction flows
• Integrates with retailer systems
• Provides a scalable foundation for agent-driven shopping and service experiences

Mirakl’s role: provide the infrastructure, connections, and operational expertise retailers need to participate in this next wave of commerce—without rebuilding from scratch.

Join the frontier

Generated by RSStT. The copyright belongs to the original author.

Source