An exclusive interview with ChatExcel's Pang Dawei: What did he do right to complete a multi-million dollar angel round of financing?

ChatExcel recently completed a nearly 10 million yuan angel round of financing, invested by Shanghai Changlei Capital and Wuhan Donghu Angel Fund.

Official sources revealed that the funds will primarily be used for product research and development iterations and global operations, further advancing its presence in the "AIdataintelligentbody(Data)AgentExploration and implementation in the direction of ").

ChatExcel as AI Native The team has served over ten million users in the past two years and has been featured on CCTV's "Winning in..."AIWinning the Entrepreneurship CompetitionintelligentThe office team achieved the best results nationwide.

Today, it can not only handle data in various formats such as Excel, databases, and web pages, but has also launched the world's first compatible...AIPCThe locally deployed vertical model constructs a complete data closed loop.

According to founder Pang Dawei, this is just the beginning: "ChatExcel is not just a tool, but also a tool for the team to keep up." AI The first ticket to this era. We hope to leverage end-to-end data capabilities to connect data acquisition to value realization, enabling ordinary people to own their own data.intelligentbody.

Product Name: ChatExcel

Slogan: Only through chatAIIt can handle Excel and data analysis.

Main function: In a single sentence, process Excel spreadsheets and database data, no functions required, and gain data insights in 10 seconds.

Official website: ChatExcel

The following is our interview with Davis, the founder of ChatExcel:

K Sister Research Society

Hi Davis, could you tell me what you've been busy with lately?

🧑🏻‍💻 Davis

Recently, the main focus has been on... AI DataAgent(AI dataintelligentbodyWe are iterating on our products in this direction while also promoting commercialization.

 K Sister Research Society

I've noticed your recent visitor data, which shows a significant increase compared to last month. Could you share how you achieved that?

🧑🏻‍💻 Davis

I think the key factor is product iteration. Our traffic increased by approximately 175%, placing us in the top five fastest-growing categories on the leaderboard. Our domestic traffic ranking is around 58th. This result exceeded our expectations.

We did almost no commercial advertising, so the growth mainly came from product iterations. Meanwhile, exposure on major platforms and channels also contributed to the increased traffic. Overall, product performance and organic exposure drove the growth.

Not because AI The reason I'm doing Excel is because Excel itself is worth doing.

 K Sister Research Society

I saw on your official website that you appeared on CCTV's "Winning in..."AI+》intelligentOur office team won first place. Could you share how you got this first place?

🧑🏻‍💻 Davis

WinningAI"+" was a program produced by the CCTV show "Win in China" back then. AI This is a startup-themed program that has been revived by the times. We first participated in the preliminary round at Alibaba Cloud's Yunqi Conference last year, which was broadcast live. After three rounds of competition, we finally entered the finals and won first place.

The competition format is similar to what you see on TV: founders need to introduce their product and business model within two or three minutes, and judges and the audience will score them.

This was a great opportunity for our team, not only giving us exposure but also prompting us to rethink: What are our core competencies? Where are our goals and barriers to entry? What direction should we take in the future? It was a bit like an "elevator pitch," requiring us to clearly explain our product and vision in a very short time.

More importantly, on a platform like CCTV, we gained a lot of traffic and exposure, but also faced sharp criticism from judges and investors. These incisive questions prompted us to reflect on where we needed to improve and how we should prepare for the future. Overall, this program was extremely helpful to us.

K Sister Research Society

What were your initial motivations for creating ChatExcel?

🧑🏻‍💻 Davis

Our core team members are from Peking University, and two of our PhD partners were already conducting research on data processing and Excel while pursuing their master's degrees. Back then, we were still using... BERT This generation AI Technology, not because of GPT We only started doing it after it appeared.

Later, during their doctoral studies, under the guidance of their supervisor, the two doctoral students attempted to re-implement data processing using the Transformer architecture. By the second half of 2022, we gradually switched to this new architecture. And just then, in 2023… GPT As the product gained popularity, we naturally positioned it as ChatExcel and launched it to users.

Therefore, our starting point was genuinely to solve Excel problems, not because of any other reason. AI This direction was determined only after it appeared.

The more difficult the problem, the more valuable it is.

 K Sister Research Society

Your team currently has fewer than 10 people. What is its composition?

🧑🏻‍💻 Davis

Our team actually only has 8 people, almost all of whom have product and technology backgrounds. You could say we are a highly technical team, and we firmly believe... AI era.

 K Sister Research Society

This is indeed a very small team size.

🧑🏻‍💻 Davis

AI The product's success relies on its technological capabilities, which is the PFG (Product Power Generation) model. Therefore, we believe that growth must depend on the iteration of product strength. Other application methods are merely icing on the cake, but they may not be able to withstand the test of technological iteration cycles.

So even though we only have 8 people, we still iterate our products in the same way we iterate our technology.

 K Sister Research Society

New architecture and AI What is the biggest difference between technology and traditional methods when processing Excel?

🧑🏻‍💻 Davis

The biggest difference lies in the change in users' habits and expectations. In the past, users needed to memorize functions, formulas, and even write code; now, users only care about the results, not the process. They expect immediate feedback as soon as they hand over tables and data to the system.

This approach actually aligns more closely with people's actual expectations and usage processes, just like using Chat.GPT Similarly, just clearly state your needs and get the result.

 K Sister Research Society

You just mentioned that your team is almost entirely composed of engineers, yet you already have a large user base in China. How did you reach so many users?

🧑🏻‍💻 Davis

It mainly relies on product strength and user-generated word-of-mouth. We don't have much money to invest in traffic or find big accounts to promote our products; growth comes from the value of the product itself.

In addition, we started relatively early, giving us a first-mover advantage. Strictly speaking, we were among the earliest groups in China to do this. AI It's an application. Although there was a period of stagnation, after two years, we are still ranked first in this field. The reason behind this is that...AI +Excel、AI Data processing is a very high-barrier field, and it's almost impossible to find comparable products on the market.

People often say "difficult but correct," and I think that makes a lot of sense. For example, in AI In this era, what we should really be addressing are the challenging questions: How to ensure accuracy? How to support multiple data types? How to handle large databases? How to ensure data security? These are all technological barriers.

The reason we've been able to accumulate so many users is because we continuously iterate and gradually meet their expectations.

We gradually built up our user base from 10 users, then 1,000, and finally 10,000. After users used the service and provided feedback, we iterated on it to meet more of their needs, which led to continued use and positive word-of-mouth.

Therefore, our growth is essentially driven by product strength.

How can we convince users that the results are accurate?

 K Sister Research Society

Many users have now... AI It's used as a daily office tool, such as for writing documents and creating PowerPoint presentations. However, in data scenarios...AI The problem of perceived accuracy still occurs, often requiring users to repeatedly verify the results. How do you address the issue of "whether the results are accurate"?

🧑🏻‍💻 Davis

That's a good question. We work with data.intelligentbodyExcel is just one type of file.

So how do we solve this accurately? There are two aspects:

First, we have behind us AI We believe in doing it through programming. AI Coding's capabilities and processing rules are accurate.

Secondly, the process is made trustworthy and visible. I make the entire chain of file and data processing visible to the user, and they can intervene. The process is transparent.

For example, when we provide services to the Big Four accounting firms, they require that the coding process be visible so that they can prove that the logic is sound.

How do you prove credibility? Your results must be verifiable.

Furthermore, the accuracy of data is also related to the data source. If the data source is of poor quality, the results will certainly be poor, so the issues of data quality and data governance also come into play.

Therefore, this issue cannot be solved solely by model capabilities; it requires a multi-dimensional approach encompassing models, product interaction, visualization, and data sources. The entire data...Agent The link is actually very long.

Commercialization direction – productization of form and productization of capabilities

 K Sister Research Society

How is your revenue doing right now? Have you started making a profit yet?

🧑🏻‍💻 Davis

We make money, we make money.

 K Sister Research Society

I see you're online. SaaS Enterprise EditionAIThe company offers private deployment versions for PC and all-in-one devices, and also provides open MCP and API interfaces. With many product forms available to the public, what is your key focus for commercialization?

🧑🏻‍💻 Davis

Our focus is really on one thing: productization. We deliver as a product, ultimately targeting end-users (C-end).

This "C" can refer to an individual user accessing our website via the internet, downloading a client, or experiencing it on other products through MCP, including our APP version and enterprise version, all of which revolve around this point.

At the same time, we are also addressing another key issue: security. DataintelligentbodyIn usage scenarios, the most pressing concerns are security and sensitivity.SimpleThe solution is to keep the files off the laptop, off the local area network, and off the company premises.

We performed quantization training on the model and created a separate data model that can support deployment on... AIIn PCs, all-in-one computers, or within a local area network.

These are our two directions for commercialization:

• Productizing the form;
• Productizing capabilities – enabling the deployment of models onto a single computer.

 K Sister Research Society

From your perspective, what would be the approximate revenue contribution ratio between C-end and B-end in the future?

🧑🏻‍💻 Davis

We primarily focus on individual consumers (C-end users), and our revenue currently comes mainly from them. While we also have some business-to-business (B-end) clients, our overall focus remains on solving data processing problems for C-end users. We don't develop traditional BI products, so our C-end customer base is currently larger.

Unlike WPS and Office, they are not comparable.

 K Sister Research Society

Currently, there are quite a few applications available for C-end users to perform spreadsheet and data processing, such as WPS and Office, which also integrate these functions. AI With existing capabilities, and newer tools like LoveCharts, how do you compete with these tools? What differentiates you?

🧑🏻‍💻 Davis

The difference lies in our positioning. We don't just do table processing; our core focus is data analysis.Agent(dataintelligentbody).

Software like WPS and Office are essentially spreadsheet editors; they just have additional features added to the editor. AIOr it can solve a single problem, such as generating charts.

ChatExcel is built around the entire data journey (data acquisition, data preparation, analysis and insight, data visualization, decision-making application, and value exchange).AI DataAgentThis allows data to flow freely, creating a closed-loop business model.

Different nodes have different scenarios, and each real user has different data needs in different work scenarios.

We do something different from WPS and Office. We consider ourselves something new, incomparable, right?

Although there is some overlap in the problems we solve, our technological approaches and positioning are different. Competition certainly exists, but I believe truly effective competition must be... AI Competition between new things of the era, rather than "old products plus". AI"This model."

 K Sister Research Society

It's equivalent to you being AI Native applications, without the obstacles of traditional products?

🧑🏻‍💻 Davis

Yes. Our greatest competitive advantage lies in the fact that we are a... AI The Native team is doing... AI Native products. Our perspective is to develop products as the model's capabilities improve and evolve with the model's maturity, which is completely different from old products.

Established products certainly have their advantages and barriers to entry, but our biggest opportunity today lies in—new products. AI What problems can the product solve?

Our team has fewer than 10 people, yet we've achieved such data and commercial success. The underlying logic is that we are... AI Native team.

AI The benefits it brings us are very obvious, whether it's our perspective on things, team composition, or the evolution of our technology. In fact, you'll find that...AI The model's capabilities are very evident in our team.

The biggest opportunities in the future are overseas.

 K Sister Research Society

How are you making progress in overseas markets?

🧑🏻‍💻 Davis

Our overseas product (English version) has been launched and we are currently promoting it internationally. However, we are still a "newcomer" in the overseas market.

 K Sister Research Society

Are the overseas and domestic versions completely separate?

🧑🏻‍💻 Davis

Yes, it's completely isolated. We used a brand new domain name. While there are some similarities in the interaction, it's a completely new name and platform overseas, so we started from scratch. PMFMarket testing and commercialization testing.

We firmly believe that the biggest opportunities in the future lie overseas. Essentially, we are a productivity tool, and overseas users are more willing and able to pay for productivity tools, so this presents a significant opportunity for us.

We are also increasing our investment, and in the second half of the year we will focus a lot of resources and energy on iterating our overseas products.

 K Sister Research Society

What are your thoughts on the differences between domestic and overseas users?

🧑🏻‍💻 Davis

Overall, we are dealing with end-users (C-end), but there are still significant differences.

First, there are differences in usage habits and payment habits. Domestic users themselves vary, for example, between first-tier cities in the southeastern coastal areas and other regions, or between different industries.

Overseas markets can be further subdivided into Europe and America, Japan, South Korea, South America, and the Middle East, where usage habits are completely different, reflecting differences in the level of information technology and payment methods.

We are still working on this, and we may have more conclusions in a few months. But based on our current experience, there are indeed differences in usage habits and payment capabilities among different countries.

However, it's not too difficult for us because the data we deal with is essentially spreadsheets: Excel, Sheet, or Data, which are globally accepted. Therefore, localization isn't particularly challenging.

An 8-person team has already secured nearly 10 million yuan in funding.

 K Sister Research Society

Can you share some updates on your recent funding rounds? Are you considering starting a new round?

🧑🏻‍💻 Davis

We completed one round of financing in the first half of the year and may launch a new round soon. We will officially announce the previous round of financing this week, which will also prepare for the next round.

K Sister Research Society

What are the main objectives of this round of financing?

🧑🏻‍💻 Davis

The funds are primarily used for product iteration and overseas markets. Although our team is already making money, expanding overseas requires significantly more investment.

In addition, at this stage, speed is key. Whether it's team depth, technological iteration, or user growth strategies, all require financial support.

 K Sister Research Society

What kind of investments do you prefer? Do you screen investments?

🧑🏻‍💻 Davis

We don't have a particularly clear inclination, after all, it's still in a relatively early stage.

The previous round of financing has been completed. Next, we hope investors will endorse our data-driven approach.intelligentbodyThis direction offers both knowledge and resource support. Especially overseas, if investors have experience in this area, it will be even more valuable to us.

Frequent iteration can help usfastFind the right direction

 K Sister Research Society

You update very frequently. How do you choose the iteration direction? How are you able to maintain such a high update frequency?

🧑🏻‍💻 Davis

This is consistent with our... AI It's related to the understanding of products in this era. Essentially, it relies on the growth of product capabilities to attract users.AI Users in this era are willing to grow together with startup teams, so we will continuously iterate according to user needs.

Secondly, frequent updates can help us.fastFind the right direction. A higher iteration frequency means a shorter trial-and-error cycle and lower trial-and-error costs, allowing us to continuously obtain positive feedback.

The shorter the positive feedback cycle, the more correct the direction.

If updates are only given every six months, the trial-and-error cycle is too long. AI The times are simply outdated. Everything changes dramatically in just one month now, so the frequency of updates is extremely important.

 K Sister Research Society

I started following you around the end of 2022, but your product development didn't gradually improve until 2025. Some people have commented that your progress has been relatively slow.

🧑🏻‍💻 Davis

Such comments are indeed common. But the underlying reasons are quite objective: our team comes from Peking University, and our two PhD partners were still pursuing their doctorates at the time, so their studies were their top priority. For example, last year they were in their first year of doctoral studies, so there was almost no iteration during that period, and their main focus was on their studies.

AI Entrepreneurs must acknowledge that they have boundaries.

 K Sister Research Society

How do you view your current niche? For example, you already have partnerships with leading companies like Alibaba and Huawei.

🧑🏻‍💻 Davis

I think a niche itself is a form of competitiveness and a barrier to entry. But the core issue is whether your product is unique within the industry and whether it can solve a specific, vertical problem—something that major companies may not have allocated resources to in the short term.

In addition, I have always believed that AI In this era, especially in China, startup teams must be very open to ecosystem collaboration.AI Technology has models and boundaries, and so do the capabilities and resources of a team. Therefore, collaboration and openness are the best growth strategies, rather than working in isolation and overlapping on the same thing.

I think today is for AI This is a great era for this generation of entrepreneurs and products. Large companies are willing to open up their resources, platforms, and traffic to help and accept small third-party teams like ours.

We need to embrace this rhythm and maintain an open mind. Actually, everyone's logic for MCP (Multi-Channel Programming) is the same now. It's already quite challenging to do a good and solid job in a specific area. We must believe that our team has its limitations; don't assume that just because your team is large, you can do everything. This is very important.

 K Sister Research Society

Domestic vendors like Huawei, Alibaba Cloud, Tencent Cloud, and ByteDance already provide server and database services. Are you worried that they might directly copy your products and offer them to their customers?

🧑🏻‍💻 Davis

We never worry about this issue. Actually, it's not just that... AI In this era, all software products face competition; this is the norm. Before we were created, many teams were already working in similar directions, and competition is always all-encompassing.

But AI In today's rapidly evolving world, we don't believe that large companies necessarily have an advantage in this area. Especially in China, large companies have higher priorities and may not necessarily invest resources in this specific niche.

For us, the most important thing is to focus on doing our own thing well. We are doing... AI For native new products, user expectations are also very high. The customer base we haven't yet reached is already very large, so instead of worrying about whether others will do it, we should focus on how and at what speed to meet user needs.

Positioning is to do AI DecisionAI decision makingintelligentbody)

 K Sister Research Society

I saw on your official website that you have two products that are about to be launched, called ChatData and ChatPanel. Could you please introduce them?

🧑🏻‍💻 Davis

These two products will be launched within two weeks. They are actually part of our product iteration roadmap. Our positioning is to... AI DecisionAI decision makingintelligentbodyThey started by focusing on high-frequency, commonly used scenarios like Excel, initially serving end-users (C-end) to accumulate data and usage habits.

We believe that data has multiple types. The first, Source, is a file type, and the second is a database type.

Our upcoming ChatData platform handles database-type data. For example, if you have MySQL, SQL Server, Spark, or other databases, we support direct access via... AI To handle it.

ChatPanel targets external data. The data generally falls into three categories: file-based data (such as Excel files), database-based data (mostly internal company data), and publicly available data, such as web-based data and third-party internet data. We will...automaticExtract data from web pages and generate tables and analysis reports.

For example, I want to know which mobile phone brands had the highest sales volume across the entire network during the 618 shopping festival, and how new energy vehicles ranked.intelligentbodyIt can directly capture and generate visual results.

From a product roadmap perspective, we've expanded from Excel to databases, then to publicly available third-party data, and even support integration with third-party APIs and the MCP protocol. Regardless of the data's source, we aim to cover the entire data lifecycle, from acquisition and processing to application analysis.

The new products mentioned on the official website are actually in line with this direction.

AI The era has just begun; ChatExcel is the starting point, not the end.

 K Sister Research Society

I understand that David was an entrepreneur before, is that right?

🧑🏻‍💻 Davis

 K Sister Research Society

Previously you worked on a marketing-focused B2B SaaS platform, and now you're working on ChatExcel. Will this be your last startup project?

🧑🏻‍💻 Davis

I consider myself a serial entrepreneur, primarily focused on marketing-related enterprise SaaS. As for whether ChatExcel will be my last project? I don't think so.

ChatExcel was just a project name at this point in time.AI The era has just begun, and I see it more as a ticket for our team and myself to enter this era.

It only means that we have just caught up with the pace of this era. There are still many things to do in the next few decades or hundreds of years, and it is impossible to predict exactly where we will end up.

Therefore, whether this is the first or the last project is not important. What truly matters is choosing the right direction in this era, persisting in it, and maintaining an open mind to continuously iterate.