Zilliz's Charles Xie: Entrepreneurship Means Confronting Your Own Ignorance Every Day | 5Y Capital Tavern Vol. 8
How do you face uncertainty when walking through no-man's-land?
5Y Capital Pub Notes
When venturing into uncharted territory, how should entrepreneurs face uncertainty?
In this edition of 5Y Capital Pub, Charles Xie, founder of Zilliz, shares his story. When he founded Zilliz in 2017, Charles chose a path few had taken — building vector databases, an entirely new category with no prior experience to draw from. Even his own team had doubts.
Every pioneer faces this predicament: embracing innovation means embracing massive uncertainty. Charles notes that the hardest part of entrepreneurship is "facing your own ignorance every single day."
But like the Guardians of the Galaxy, exploring the vast universe is itself the mission for Charles and his team. When you choose to see problems through a wider lens and from a higher dimension, uncertainty itself becomes the fuel that drives you forward.
Perhaps his story will offer you some inspiration :)
Our guest for this edition:
Charles Xie, Founder & CEO of Zilliz
About Zilliz:
Zilliz created Milvus, the world's most popular open-source vector database, and is a pioneer and global leader in vector database systems. Zilliz's technology and products are used by over 1,000 enterprises worldwide, with applications spanning computer vision, image retrieval, video analysis, natural language processing, recommendation systems, targeted advertising, personalized search, intelligent customer service, fraud detection, cybersecurity, and new drug discovery.
What we discussed:
- Charles's next destination is always the vast universe.
- Buying 100 identical pieces of clothing after starting a company.
- History cannot be replicated at the micro level, but follows patterns at the macro level.
- The hardest part is facing your own ignorance every day.
The Next Destination Is Always the Vast Universe.
5Y Capital Pub: Why do you call yourself Star-Lord? When did you start using that name?
Charles Xie: It's a company tradition. Every new hire's first task is to choose their own hero — it can be a real historical figure or a fictional character from any time or universe.
Why Star-Lord? The Guardians of the Galaxy are among the few Marvel superhero teams that operate as a group. Individually, each hero may not be that powerful, but together they form an invincible team — and they keep growing stronger over time. I deeply admire this team spirit, accomplishing seemingly impossible tasks through collaboration.
Also, most other Marvel heroes have a home planet, a hometown. But Star-Lord and his Guardians have made exploring the vast universe their mission. After completing one mission, their next destination is always the endless cosmos. They remain perpetually curious about the unknown. This is a spirit I personally admire, and I hope I can always maintain this curiosity.
5Y Capital Pub: How did this tradition form, and what hero choices from others in the company have stuck with you?
Charles Xie: I proposed it initially, and after some discussion we found it fun and decided to make it part of our culture. When each person chooses their hero, they explain why. What's interesting is that many people choose heroes that somehow connect to or resemble themselves. For example, our very first engineer chose Groot. Groot in the Guardians of the Galaxy is a character who really steps up — dedicated to the team, willing to contribute, even sacrifice himself. This engineer demonstrated exactly these qualities in his actual work.
5Y Capital Pub: Has your understanding of Star-Lord changed over these years of entrepreneurship? How do you see this hero now, and have you found any new heroes?
Charles Xie: Not really. Every hero has flaws; the difference lies in their particular traits. Perhaps it's like people — gradually recognizing your own shortcomings and deficiencies, constantly striving to improve.
Star-Lord is actually a very human hero. He's not高高在上 or cold and distant; you can sense a strong everyman quality about him. Over the past few years, I've increasingly felt that entrepreneurship requires this trait too. After all, entrepreneurship isn't about performing under spotlights for a few hours. It's built from every day, every detail of life.
5Y Capital Pub: What has your daily life been like since starting your company? You mentioned buying 100 identical black shirts after founding Zilliz to save time and mental energy. Has anything changed?
Charles Xie: Yes, still the same, though not just black anymore — I'll buy dozens of the same style of T-shirt in one go. I often encounter this scene at checkout: the salesperson looks at me with concern and gently reminds me, "These clothes aren't on sale today."
I've maintained this simple approach to dressing since starting my company. This way I don't have to think about what to wear each morning, saving that mental energy for more important and critical problems.
Being busy is the norm in entrepreneurship, with increasingly high-dimensional challenges constantly emerging. But entrepreneurship is a marathon, so taking time to recharge, think deeply, and maintain good condition is essential. When someone is completely overloaded, they may make foolish decisions. Regardless of my state, I always reserve some time for learning and relaxation — reading, listening to music, watching films, or exercising.
History Follows Patterns at the Macro Level.
5Y Capital Pub: When you first built this innovative product, did you encounter skepticism or misunderstanding from the market? What was that process like?
Charles Xie: When I founded Zilliz in 2017, I was determined to build a new database system for the AI era. At the time, the market was basically full of voices that didn't understand. The vast majority of investors didn't grasp why the AI era needed entirely new data infrastructure software. Most pitches got roughly the same feedback: "That's interesting." Fundraising was extremely difficult for a time.
In November 2017, I had a chance encounter with 5Y Capital. I still remember clearly — it was 11 a.m. on November 16. I was invited to Shenzhen to meet with 5Y. We discussed then-trending AI and big data technologies, and also my vision for future data processing in the AI era. The meeting was scheduled for one hour, but everyone forgot about lunch, and we discussed intensely for nearly two and a half hours on empty stomachs. Within the following week, 5Y led our Series A round.
Looking back, as 5Y's slogan goes — "When others see crazy in you, we start to believe" — 5Y's investors are themselves seen as crazy by others. Only crazy people support crazy ideas, and only they can accompany crazy founders all the way to the day they start being believed.
When determining our R&D direction, our internal engineers were also full of concerns about this entirely new category, with some hesitation and debate. Initially, we saw new technologies emerging in the AI era, along with distributed computing and heterogeneous computing capabilities. But what form the combination of new AI technology and new data processing capabilities would ultimately take as a product was very unclear at the start.
We were embracing innovation while also embracing massive uncertainty, trying to distill common patterns from behind this uncertainty and behind the market's diverse demands — could we build an entirely new database infrastructure product for the AI era?
It took nearly two years of constant exchange and collision with the market and users. Finally in 2019, we focused on the vector database category. At the time, we were the first in the world to create this category. When we open-sourced the product in September 2019, it gained recognition from dozens of enterprise users within months. We felt the product direction was right, that it genuinely solved users' pain points. From there we focused on vector databases and have continued to this day, watching many vendors join the innovation in this category. We're quite proud to have pioneered a new category.
5Y Capital Pub: Do you feel anxious facing this kind of massive uncertainty, and how do you personally deal with it?
Charles Xie: From the perspective of broader technological history, this feels relatively certain to me. Looking across the entire history of human technological civilization, you can see larger trends that transcend cycles.
I've always believed that history cannot be replicated at the micro level — as philosophers say, you cannot step into the same river twice. But history follows patterns at the macro level, even repeating itself, so we can find some certainty within a more macroscopic dimension.
Before entrepreneurship, I had worked in databases for about 15 years. Over the past decade-plus, database products have undergone a trend of socialized division of labor. The field went from a few general-purpose relational databases to a blooming of diverse vertical databases — document databases like MongoDB, graph databases like Neo4j, time-series databases like InfluxDB, and so on. I observed this trend and firmly believed that database development, like many fields in human history, would further divide as user needs became more granular. In the AI era, entirely new database systems would emerge, such as vector databases; in the quantum computing era, quantum databases might even appear.
Additionally, our vector database was open-source from day one. Over the years we've continuously received user feedback. With each product iteration and new feature launch, we could validate hypotheses through user feedback. Many hypotheses proved correct; many proved deviated from reality, and we would promptly make adjustments in subsequent versions.
These two approaches have helped greatly in alleviating anxiety from uncertainty: looking at problems through historical cycles, and getting feedback from frontline users to rapidly test hypotheses.
5Y Capital Pub: Were there any difficult decisions you faced along the way?
Charles Xie: In the second half of 2020, our vector database Milvus version 1.0 had achieved good results, with about 300-400 users globally. The project initially adopted a relatively simple single-machine architecture, providing efficient processing for tens of millions of vector data points. The benefit was rapid time-to-market. But over the past few years we saw vector data growing rapidly, and predicted potential tenfold or even hundredfold growth in the future. In this scenario, single-machine architecture had limited scalability, making it difficult to meet future needs for processing billions or even tens of billions of vector data points.
By the end of 2020, we faced a choice. If we continued with the 1.0 architecture, we might hit a major scalability bottleneck in some future year. But if we redesigned from scratch, it would mean starting over, rebuilding the product in a more scalable, cloud-native way.
This was an especially difficult decision. At the time, our 1.0 architecture was still quite popular. Moving to a new architecture introduced much uncertainty for product and R&D timelines, and for a startup, this meant significant time and economic costs. Looking back even now, this was a very difficult decision because it could have meant the survival or death of the company.
5Y Capital Pub: How did you ultimately make this decision, and were there disagreements along the way?
Charles Xie: It took about a quarter to decide. We had countless meetings, with much argumentation and analysis of different approaches, trying to surface the uncertainties to see if we could solve them. Through this process, everyone gradually aligned.
What ultimately drove our decision was a rather pure engineering philosophy: as engineers, we should build a more technically advanced product solution that serves users better.
In the second half of 2020, we began rebuilding our 2.0 project. In retrospect, this decision was well worth it. Because of our early layout and R&D, our product has massive advantages in scalability and cloud-native capabilities in the market. Today we're the first vector database startup globally capable of providing billion-scale vector retrieval.
5Y Capital Pub: Zilliz chose open source from the very beginning. Why this direction? When did you first encounter open source, and what's your story with it?
Charles Xie: Around 2000, when I was still in university, I encountered the open-source web server project Apache. This was my first exposure to open source. Later during graduate school, I had the opportunity to participate in open-source software development. There was a famous open-source project in grid computing called Globus, developed collaboratively by the University of Chicago and Argonne National Laboratory in the U.S. We participated in this project's collaboration. Because of open source, a team in Wuhan, Hubei, China could collaborate with teams in Illinois, USA, keeping technically synchronized — this was exciting. And because of open source, we could access the latest global technologies and participate in their iteration ourselves.
Later in Zilliz's founding process, one of our core cultural values was staying committed to open source and openness. This is an important principle for our company, and also aligns with the essence of programmer spirit — "Talk is cheap. Show me the code." We also hope that through open source, new technologies and products can reach different industries earlier and faster, helping them create value. A company's value ceiling ultimately depends on how much value it can create for users.
The Hardest Part Is Facing Your Own Ignorance Every Day.
5Y Capital Pub: What do you find most difficult in entrepreneurship — hiring, organization, strategic direction? What's the hardest?
Charles Xie: It's not those things. The hardest part since starting my company is facing my own ignorance every single day. You may pay for yesterday's ignorance today, and use today's ignorant self to face tomorrow's uncertainty.
This has also been our team's greatest challenge along the way. There's uncertainty at every stage of company development; at different stages and scales, organizational building, people management, and culture are completely different. Additionally, because we're doing something no one has done before, walking in uncharted territory, we must constantly face our own ignorance and upgrade our cognition.
5Y Capital Pub: What's your personal learning approach? How do you overcome the mindset of "facing your own ignorance every day"?
Charles Xie: First, admit your ignorance and face it directly. There are several things I've consistently done. First, I find good role models and more experienced people to learn from through exchange, making my ignorance more concrete — so I know where my misconceptions and blind spots lie.
After finding the misconceptions, I engage in extensive learning, finding excellent books in that field to study. The final stage is also particularly important: after learning through reading, I meditate. Through meditation, I can absorb and integrate knowledge, making it part of my own cognition.
These three stages form a closed loop of cognitive upgrading. After completing one cycle, I can start the next — finding more experienced people in different fields to talk with. For me, I generally try to complete one full major iteration each year.
5Y Capital Pub: Through these cognitive upgrades, have your views on the world changed significantly in recent years?
Charles Xie: Perhaps the biggest change in recent years is that I believe human cognition has gradients, or different dimensions. A higher-dimensional thinker looking at lower-dimensional problems finds them simple, quickly arriving at solutions.
Like when we first studied calculus in college — suddenly the hardest analytic geometry problems from high school became easy to solve with calculus. Later as an engineer using programming to solve many problems, you might find that with more advanced data structures and design patterns, many code problems encountered at lower dimensions are readily resolved.
Whether in entrepreneurship or life, what needs solving is ultimately cognition of the entire world. Many things that seem hopeless and completely unsolvable at lower dimensions may be handled with ease from a higher dimension.
For me personally, when I encounter great difficulties, I comfort myself: it's because your current thinking dimension is too low. Elon Musk is thinking about going to Mars, such a difficult problem, and you're still tearing your hair over early-stage daily operations. Of course I know Musk also gradually elevated his dimension step by step, but I use this to give myself confidence. The answer exists; it may just take more time, looking from outside the space. As we say, think out of the box — jumping out of the box to look at the problem is essentially switching to a higher dimension.
5Y Capital Pub: But this dimension-elevating process may also be painful, because it's hard to have higher-dimensional thinking while still inside the lower dimension.
Charles Xie: I find it quite interesting that building higher-dimensional cognition is actually quite anti-human-nature. As people age, their thinking may also rigidify. You must break existing cognition to reach higher dimensions. Existing cognition, past successes, and honors may all become stumbling blocks to reaching the next stage.
So you must recognize your ignorance, then have the courage to break it — this process is quite anti-human-nature, and quite painful.
5Y Capital Pub: You may not know when you can enter the next stage, or even whether you can complete it. How do you face this uncertainty in timing?
Charles Xie: Facing this uncertainty may itself be the source of each individual's pursuit of self-actualization. If someone told you that at 40 you'll reach a certain stage, at 60 another stage, with no uncertainty in between, perhaps no one would want to strive anymore — there'd be no challenge.
I think the reason many people take self-actualization as their ultimate goal, striving one after another, is fundamentally that this uncertainty attracts them. Like the line from Forrest Gump: life is like a box of chocolates — you never know what you're gonna get next. This uncertainty also gives us motivation.
5Y Capital Pub: What's your ultimate goal for self-actualization?
Charles Xie: I hope that on the day I bid farewell to this world, I can still hold this mindset: honestly accepting my ignorance, and exploring this world.
Maintaining curiosity about the world is actually very difficult. Many great figures in human history may have fallen into mental stagnation and rigidity in their later years. I hope to make this a lifelong requirement for myself.
Interactive Gift
Welcome to share your views in the comments. We'll select 3 featured comments to receive a 5Y Capital commemorative T-shirt + 5Y Capital coffee cup. (Comments accepted until October 21. Please reply with shipping information within 24 hours of receiving notification.)
5Y Capital seeks out, supports, and inspires lonely entrepreneurs, providing support from spirit to all operational matters. We believe that if others see crazy in you, and we start to believe, the world will become a different place.
BEIJING · SHANGHAI · SHENZHEN · HONG KONG
