Hillhouse Capital and Hongshan Co-Invest in a Sales Agent Startup Going Global
Former Youzan COO Launches New Venture
"Former Youzan COO's Second Act"
By Xu Muxin
Edited by Chen Zhiyan
An Yong Waves has learned that Dealism, an AI Agent company founded by former Hillhouse Capital investor and former Youzan COO Huan Fang (Leo), has completed a $15 million angel round led by Hillhouse Venture Capital, with Hongshan and Linear Capital participating. This round combines two tranches into a single close. Beyond institutional investors, the round includes the founder's personal capital and multiple individual investors.
Leo's career began at Hillhouse Capital. From 2011 to 2018, he served as VP there, investing in companies including Youzan, Zoom, and NIO. In 2018, he joined Youzan as co-president and COO, spending seven years overseeing commercialization, sales systems, and growth strategy. Now he's chosen to start again, with a mission to "use AI to replace the massive volume of repetitive, inefficient communication tasks in sales roles."
Dealism was formally established in July this year.
The company's core product is a conversational sales Agent built around the "Vibe Selling Agent" concept — an AI sales partner that anyone can pick up and use, one that reads customer psychology, detects emotional signals, and pushes deals toward close. It's targeting Latin American and North American markets, serving SMB customers including independent creators, local service providers, and e-commerce sellers. The team currently numbers under 20, headquartered in Singapore with plans for global expansion.
On the eve of the product launch, An Yong Waves sat down with Leo. He explained why he's gone all-in on "vertical Agents" and why he believes this year represents the best timing yet for Agent entrepreneurship. "Because the best time to plant a tree is today."
The conversation follows —
Part 01
Not a Chatbot, but a Sales Avatar
An Yong: What's the difference between a "Vibe Selling Agent" and an intelligent customer service bot?
Leo: We're not building a chatbot in the traditional sense. "Chatbot" is an extremely broad concept — the truth is, there's no such job as "chatting" in this world. What we're actually building is an object-based, purpose-driven conversational Agent. It understands who you are and what you're trying to accomplish, then autonomously communicates with external customers to achieve that goal.
Think of it as a personalized conversational Agent for every individual. It can learn from your materials, your communication style, your sales process. And this Agent has continuous learning and memory capabilities — it gets to know you better over extended use, and keeps improving on more "human-like" dimensions like emotional intelligence, customer psychology, and negotiation.
An Yong: So you're essentially giving every salesperson an "AI avatar"?
Leo: You could put it that way. I also want to make it easy for ordinary people to pick up and use — including all kinds of solo entrepreneurs and side-hustlers — so they can achieve their goals with a sales Agent.
An Yong: So your target customers are overseas "Xianyu sellers"?
Leo: Our primary markets are North America and Latin America. Broadly speaking, anyone in a sales or marketing function could become our user. The tier skews toward Pro C and SMB — for example, Chinese brands going global and cross-border traders.
An Yong: Why do they all need a personal sales Agent?
Leo: They're the classic "high customer service pressure, but not professional at it" demographic. First, compared to China, their overall degree of "online-ification" is much lower — though the shift is happening fast — and labor is expensive. Chinese e-commerce demands 30-second response times, but if you're running an overseas independent site, having someone respond within 24 hours is already pretty good, let alone for influencers with massive followings. So against this backdrop of huge communication demand, when you can't afford a professional operations team and don't have an existing CRM system, the Agent can directly fill that gap.
An Yong: How is Dealism thinking about monetization right now?
Leo: Currently it's per-user subscription plus session volume. We'll transition to outcome-based pricing in the future.
An Yong: How long until "delivery can be measured by outcomes"?
Leo: Once what I'm delivering can be measured by outcomes, the latter model becomes possible. Of course it depends on scenario complexity. I can see that in certain specific industries and scenarios, we're already very close to direct transaction closure. I'm quite optimistic on the progress here.
An Yong: You used to be an investor — "what's the endgame" is a question this crowd loves to ask. What's Dealism's endgame?
Leo: In an era of rapid large model iteration, forget endgame — judgments beyond six months carry real risk. But we can look at some relatively certain factors. First, the ceiling on this market is extremely high. There are 150 million sales professionals globally, with $3–5 trillion in annual labor market expenditure. If Agents can replace even a portion of that, it's already a massive market. So this isn't traditional SaaS — it's a new form of "personal productivity tool." What we're doing now is providing frontline productivity. It doesn't compete with traditional SaaS in this domain. I believe Sales Agent will become the sales production and management tool of the new era.
We want every sales individual globally to have their own Agent and become a super-individual of the new era. This will happen within the next ten years, and we aim to be among the first to make it real.
Part 02
Vertical AI Agent Will Be the New Productivity Form of the Future
An Yong: When did you decide to go all-in on the Agent direction?
Leo: I started paying attention to large model technology and transformation in 2022. My first core judgment at that time was: vertical AI Agent will become the new productivity form of the future.
We've all lived through the mobile internet era. That wave of technology enabled every workflow to be recorded, digitized, and thus more standardized (SOP-ified). But it didn't fundamentally change the workflows themselves — it just gave existing processes greater scale effects, which is why we saw companies with massive organizational scale emerge globally.
But this time, the intelligence and reasoning capabilities that AI brings are elements never before seen in human history of tool use. They will cause positions and workflows across every industry to undergo "first deconstruction, then reconstruction." The ultimate change: organizational scale and headcount demand will drop dramatically — possibly even below pre-internet levels — while team productivity rises significantly. The gap between these two is the more efficient, more intelligent Agent. I believe this is an inevitable trend over the next decade.
I believe the next generation of advanced organizations will be extremely small in scale. They won't need HR to help manage large organizations — they'll need AR (Agent Resource), people who introduce, deploy, and manage Agents for the team.
An Yong: Why did you judge that you're the one who can do this?
Leo: I often tell the team: we may be the team most knowledgeable about consumer goods among all AI teams. Over the past seven or eight years, while managing Youzan as a public company, I served hundreds of thousands of Chinese and overseas B2C consumer brands, covering virtually every industry and category, facilitating over 100 billion RMB in GMV annually. China is the market with the richest supply globally, which has also spawned the most advanced sales models. We've done extensive research and knowledge extraction — this is the competitive advantage of Chinese brands going global, and it's why I want to use Agent to package China's leading sales productivity and take it global.
An Yong: Did pain points from the Youzan era catalyze your entrepreneurship?
Leo: A core part of my work was managing a market and sales team of thousands, spending over $100 million in budget annually. In reality, 60% went to frontline team recruitment, training, and coaching. I hired one excellent colleague after another, who achieved great results through their own diligence and effort. But many of them left because they couldn't sustain the repetitive customer development work. The next-generation AI Sales Agent can take over this repetitive communication, freeing people to spend time on truly valuable customer connection and trust-building.
An Yong: From the current vantage point, what's the biggest challenge in achieving this?
Leo: Conversational sales is a decentralized knowledge accumulation — every industry's vertical scenario has specific practices and personalized best practices. Building an entirely new product interaction form that lets users give feedback during use, accumulating experiential data, and letting your Agent continuously learn to become more like you — this is crucial. At the same time, the product's ability to accumulate user context and buyer behavior across the full transaction chain is equally important. Buying and selling is a closed-loop behavior with very strong outcome signals, naturally suited for reinforcement learning training, and will ultimately self-reinforce to become more outcome-oriented.
An Yong: Agent entrepreneurship is hot this year, and crowded. What do you think it takes to win?
Leo: Agents can do many things. If we're talking specifically about productivity scenarios, I believe "deep industry knowledge" + "building the next-generation product interaction form" + "moving fast" are the most important.
An Yong: What do you think Hillhouse, Hongshan, and Linear's investors are fundamentally betting on in Dealism?
Leo: You'd have to ask the investors — though I'm deeply grateful for their trust. I believe vertical Agent as a new productivity form is a sufficiently large opportunity. My team and I have deep accumulated expertise here, plus the experience and strong judgment to bring the latest technology to broad users through product. On the team front, we'd love to bring on more partners with local market expansion and user growth experience in Latin America and North America.
An Yong: Is 2025 the best vintage year for Agent entrepreneurship?
Leo: There are two most important things in entrepreneurship: first, timing; second, don't think too small — the latter is harder. I believe Agent replacing productivity is a ten-year cycle, a ten-trillion-dollar opportunity, so it's hard for me to say whether 2025 is the best year. But it's never too late to act now.
An Yong: The first year of the next ten years is the best year?
Leo: Of course.
Image source: Unsplash
Recommended Reading
The flow of capital, the rise and fall of people