内容来源：https://techcrunch.com/2025/11/26/onton-raises-7-5m-to-expand-its-ai-powered-shopping-site-beyond-furniture/

内容总结：

近期，科技行业正积极将人工智能技术应用于电商购物领域。除OpenAI、谷歌和亚马逊等科技巨头持续加码AI购物助手外，一批初创企业也在该赛道崭露头角。其中，智能家具购物平台Onton（原名Deft）宣布完成750万美元（约合5400万元人民币）新一轮融资，由Footwork领投，累计融资额已达1000万美元。

值得关注的是，该平台用户规模呈现爆发式增长：月活跃用户从5万激增至200万，累计完成数百万次搜索与图像生成任务。公司联合创始人扎克·哈德逊表示，尽管大语言模型在意图推测方面表现优异，但尚未有效解决电商领域的核心痛点。据其观察，消费者平均决策时长正在持续增加。

为提升用户体验，Onton采用神经符号架构技术，有效规避了大语言模型的“幻觉”问题。其AI系统能自主学习商品未标注的实体特征，例如通过材质分析自动识别适合宠物家庭使用的耐抓挠家具。平台还创新性地融合了多模态交互方式，支持用户上传空间照片进行虚拟布置，或通过图文结合方式生成个性化设计方案。

据悉，新融资将用于拓展服装和消费电子等新品类。目前公司团队已从2023年的3人扩充至10人，计划继续招募工程师与研究人员，将团队规模扩大至15人。在服装赛道，Onton将直面Daydream、Aesthetic等竞争对手的挑战。数据显示，其创新模式使客户转化率达到传统电商的3至5倍，展现出AI技术重塑购物体验的商业潜力。

中文翻译：

大型科技公司正积极将人工智能应用于购物领域，而不仅限于内容生成与摘要撰写。OpenAI、谷歌和亚马逊已投入重金开发AI助手，这些工具能帮用户研究新品类别并推荐合适商品。

Perplexity、Daydream和Cherry等初创公司也围绕AI商品发现展开业务。这些举措促使消费者更频繁地借助AI进行购物。智能家具购物平台Onton（前身为Deft）表示，其月活跃用户数已从5万激增至200余万，累计处理数百万次搜索与图像生成需求。

基于业务增长态势，这家初创企业今日宣布获得750万美元新一轮融资。本轮由Footwork领投，Liquid 2、Parable Ventures及43等机构跟投，使其融资总额达到约1000万美元。公司计划将资金用于拓展服装品类，并逐步进军消费电子领域。

谈及年初更名原因，公司表示原名称“Deft”易引发混淆且难以获取优质域名。

联合创始人Zach Hudson指出，虽然大语言模型擅长推测潜在意图，但尚未解决电商领域的诸多痛点。他透露平台监测到消费者平均决策时长正在增加。

核心技术方面，公司采用神经符号架构。Hudson称该方案能消除大语言模型的幻觉问题，提供更精准的逻辑搜索结果，并补充说明其模型可学习现实世界中未必包含于商品描述的信息。

以宠物友好型家具搜索为例，他解释道：“我们的系统能识别含聚酯纤维材质的商品更防污耐刮，因此更适合宠物家庭。通过持续学习海量搜索数据，工具正以指数级速度变得更智能。”该AI模型还会综合考虑不同网站对同一商品的不同称谓，优化展示结果。

为辅助用户长短周期决策，Onton新增多种输入方式：用户可上传空间图片或输入需求描述，平台据此推荐匹配家具；还推出支持图像生成的无限画布功能，允许用户混搭现有图片与商品进行创意构思，甚至直接上传房间照片获取整体软装方案。

公司认为，相较于纯聊天模式，这些功能能为消费者提供更多元的探索路径——即使他们无法精准描述需求。据称，基于这些创新，平台转化率达到传统电商的3-5倍，这得益于用户对底层数据可靠性的认可。

Hudson强调，技术架构与交互界面的升级使服装品类拓展更具优势。公司正在构建该品类商品库，即将正式上线，届时将与Daydream、Aesthetic、Style.ai等企业展开竞争。

团队规模已从2023年的3名全职员工扩展至10人，公司计划继续招募工程师与研究人员，将团队扩充至15人。

英文来源：

Major tech companies aren’t just using AI to help you generate or summarize content — they also want you to use it for shopping. OpenAI, Google, and Amazon have heavily invested in AI assistants that research new product categories for you and suggest the right ones to buy.
Startups like Perplexity, Daydream, and Cherry have also built businesses around AI for product discovery. All these efforts have resulted in customers using more AI for shopping. Onton (previously known as Deft), an AI-powered furniture shopping platform, says it has seen its user base grow from 50,000 monthly active users to over 2 million monthly active users, serving millions of searches and image generations.
Fueled by this growth, the startup announced today that it has raised $7.5 million in a new funding round led by Footwork, with participation from Liquid 2, Parable Ventures, and 43, among others. This round brings the startup’s total funding to approximately $10 million.
Using this funding, the company wants to expand into new categories such as apparel and then eventually consumer electronics.
The company rebranded from Deft to Onton earlier this year, citing confusion around the original name and difficulty securing a premium domain.
Zach Hudson, co-founder of Onton, says that while large language models (LLMs) are good at guessing probable intent, they have not solved many problems in e-commerce. He added that the startup has observed that the average time a consumer takes to make a purchase decision has increased.
For its core technology, the company uses what’s called neuro-symbolic architecture. Hudson said that with this approach, the company can eliminate the hallucination problems of LLMs and provide better, logical search results. He added that the startup’s model can also learn information from the real world that might not necessarily be included in a product description.
Join the Disrupt 2026 Waitlist
Add yourself to the Disrupt 2026 waitlist to be first in line when Early Bird tickets drop. Past Disrupts have brought Google Cloud, Netflix, Microsoft, Box, Phia, a16z, ElevenLabs, Wayve, Hugging Face, Elad Gil, and Vinod Khosla to the stages — part of 250+ industry leaders driving 200+ sessions built to fuel your growth and sharpen your edge. Plus, meet the hundreds of startups innovating across every sector.
Join the Disrupt 2026 Waitlist
Add yourself to the Disrupt 2026 waitlist to be first in line when Early Bird tickets drop. Past Disrupts have brought Google Cloud, Netflix, Microsoft, Box, Phia, a16z, ElevenLabs, Wayve, Hugging Face, Elad Gil, and Vinod Khosla to the stages — part of 250+ industry leaders driving 200+ sessions built to fuel your growth and sharpen your edge. Plus, meet the hundreds of startups innovating across every sector.
“Let’s say you are looking for furniture that is pet-friendly. Our tools know that if the item has polyester in it, it would be more stain and scratch-resistant, so it would be more pet-friendly. Our tools learn these things through every single search and become smarter at a faster rate,” Hudson said.
He added that often, when you search for a product that might be called different things on different sites, you don’t get great results. The company’s AI model takes those scenarios into consideration while presenting results.
Onton has added different input methods and features to help people with their short and long-term decisions. You can now upload an image or add a prompt to generate what you want to achieve with your house or office setup, and Onton can find you furniture based on that.
Onton also offers an infinite canvas with image generation, where you can add existing images along with the products you find for ideation. You can also add images of your room and ask the tool to furnish it.
The company feels that rather than stick to a chat-only approach, these features will give consumers more options to get to what they want, even if they don’t know how to describe it perfectly.
The startup said that with these approaches, it has been able to convert customers 3-5x more than traditional e-commerce sites, as they can trust the underlying data.
Hudson noted that because of the technological and interface changes it made, it will be easier to launch apparel. The company is building its catalog for the category and plans to launch the vertical soon. In this category, it will face competition from companies like Daydream, Aesthetic, and Style.ai.
The company has grown from three full-time employees in 2023 to 10 now, with plans to expand the team to 15 people by hiring engineers and researchers.