º¬Ðß²ÝÉçÇø

In a fiercely competitive AI boom, how should Chinese legaltech companies make a name for themselves? Luna Jin talks to startup MetaSota COO Wang Yiwei about his philosophy of building products and unique promotional attempts on social media

Started as a boutique legaltech firm with 50 members, MetaSota has so far launched four popular productivity tools covering translation, writing, case search and other tasks, and has drawn wide acclaim from the legal industry.

Among these tools is Metaso.cn, China¡¯s answer to US-based AI search engine Perplexity. It is an offering that is more than holding its own in the legal sphere but also venturing into the rough-and-tumble of the general AI field, garnering public users¡¯ recognition beyond the professional domain.

Currently, the AI tech startup has achieved break even. It recently completed a new round of financing for more than RMB100 million (USD14 million), led by Ant Group, and a post-investment valuation of USD150 million. According to website traffic monitoring tool Similar Web, Metaso.cn attracted 15 million visits in June 2024, ranking it in the top five of similar products in China.

Behind MetaSota¡¯s rising reputation is a set of unique corporate communication strategies explored by Wang Yiwei, the firm¡¯s COO and a former lawyer. Rather than under the company¡¯s name, Wang has opted to build a Bilibili account under his personal identity.

Named ¡°Wang Yi Kuai¡±, it has attracted more than 600,000 subscribers. Having posted more than 300 videos, his content focuses not on direct promotion of the company¡¯s products but on self-growth and practical workplace tips for up-and-coming legal professionals. One of his videos, on how to host a business dinner, garnered more than 3 million views.

China Business Law Journal: Tell us about your background and your role as COO at MetaSota, as well as how your company has evolved and positioned itself in the legaltech industry.

Wang Yiwei: I have a background in law and have not studied any other related fields such as management or media. After obtaining my undergraduate degree from Peking University Law School, I pursued two LLM degrees at University College London and Columbia University. My professional journey started in 2006, when I joined Fangda Partners, and it continued until I co-founded MetaSota.

Around the second half of 2017, we began laying the groundwork for MetaSota. At that time, the legaltech industry was in a relatively stagnant phase, with some database products and O2O [online to offline] experiments being attempted.

It was during this period that I started to realise the potential of AI technology. The purpose of AI is to solve problems and apply them directly to legal work, not to link information or remove information barriers. This concept fascinated me. It was then that I met my partner, AI scientist Min Kerui, and we began to explore how AI technology could be applied to the practical work of lawyers.

Through our exploration, we discovered that the implementation of AI in legal contexts might progress faster and solve more problems than in other fields, which challenged my existing perceptions.

This is because legal documents are highly structured, logical, of high quality, and basically have a fixed format and generally follow fixed formats with changes following predictable logic, like contracts or IPO prospectuses.

This means that the quality of the knowledge used for machine learning in this area is very high. Therefore, it is actually easier to require an AI to generate legal content than to engage in fanciful creativity such as writing a high-quality novel.

We came to a realisation that measuring the ease of AI implementation in a particular field is not based on the depth or the amount of knowledge in that field.

In 2018, we founded MetaSota, and since then we have gradually started a series of attempts to bring AI technology to life in the legal industry. As the COO of a legaltech company, my primary responsibility is to drive traffic to the company, to make more people aware, [to] understand and use our AI products.

As for MetaSota¡¯s positioning, we¡¯re a legaltech company but we¡¯ve also got one leg in the general purpose track. The two routes, legal and general purpose, go hand in hand, and that¡¯s the way we usually think about making products.

The visibility of our AI search engine, Metaso.cn, launched at the end of February this year, is definitely higher than our other legal products. It is a general purpose product that is comparable to foreign ones like Perplexity, usable by anyone rather than specifically targeting legal professionals.

However, just before introducing Metaso.cn, in late November of last year, we launched a product called MetaLaw specifically designed for lawyers. One could say MetaLaw is the lawyer¡¯s version of Metaso.cn, with both products sharing the same technological foundation.

If a technology or application can win over the most discerning, results-demanding and hard to please groups, then we can confidently roll it out to a wider audience.

CBLJ: You have amassed more than 600,000 followers on Bilibili, a remarkable achievement. Has all this investment in social media translated into user growth for MetaSota?

Wang: Initially, we opted to advertise on social media platforms like Bilibili, Xiaohongshu and Douyin to achieve the best cost-effectiveness. However, understanding how advertising works and its patterns was crucial. Since neither myself nor my team had any background in advertising operations, to rapidly gain experience I [built] an account from scratch myself to grasp how accounts grow, and what kind of content attracts which audience. This was the genesis of the ¡°Wang Yi Kuai¡± account. Through this process, our entire team also amassed a wealth of experience in advertising.

The timing of creating the account itself seemed almost like a stroke of luck. In our previous attempts, we had explored traditional promotional methods such as attending conferences, offline events, visits and talks. Unfortunately, with the covid outbreak, all offline efforts had to come to a halt, and we had to seek the path of online development.

Starting an account on any social platform for a startup without any brand endorsement and very low national recognition is extremely challenging. My experience tells me that it¡¯s crucial for people to get to know the founder personally to form a concrete impression of the company.

I once wrote an article encouraging all founders of small companies to manage their own social media accounts. Only through firsthand experience, rather than delegating to employees, can you deeply grasp the lessons and details involved. This is also an important means of toughening a founder¡¯s resolve, and can even be seen as a required course ¨C how to deal with the public and media.

It¡¯s quite difficult for legal professionals to make our names known outside of our community. Lawyers typically interact within a niche of high-end clients and peer attorneys. While we are close to the business world, we are often very distant from the general public.

Operating in social media has exposed me to the diversity of the world, enabling a deeper understanding of the thoughts of young people and fans, a group we wouldn¡¯t typically engage with. This helps founders better understand the real world.

Founders often find themselves puzzled by these questions: Why don¡¯t people like my product? What do they like? What products are they paying attention to? Through my experience with social media, I can effectively obtain such information.

The experience can be highly beneficial. Online influence extends offline, bringing many opportunities where fans and viewers are more inclined to trust our products. During communications with various local lawyers¡¯ associations, many individuals invited me to participate in collaborations and events because they learned about MetaSota through my online visibility. Without such an experience, these opportunities might not have been as efficiently realised.

Social media operation is far from the smooth sailing one might imagine; it truly tests a person¡¯s abilities and temperament.

CBLJ: In comparison with other legaltech products on the market, what do you see as MetaSota¡¯s competitive edge?

Wang: Our strength lies in the relentless pursuit of exceptional user experiences, a core goal pursued by almost all AI products today. Besides OpenAI, very few companies dare to claim a technological edge.

User experience is a comprehensive and intuitive metric ¨C do users like the product? How willing are they to pay for it? This is the primary focus of our company.

We emphasise rapid enhancement of product experiences. When a product reaches 80/100 points in technology and production, we launch it to the market for users to use and experience, then swiftly iterate. Within a short month, we can optimise the product to a 90-point level.

Subsequently, we attempt to monetise it, observing user willingness to pay. Positive feedback prompts us to further drive marketisation and commercialisation, scaling product promotion. This has become the standard process and steps in our product development.

From the debut of our first product, Mita Translation, to Dialogue Writing Cat, and now MetaLaw, all three currently paid products follow this route. We have accumulated rich experience in this regard, and the same is true for Metaso.cn.

Currently, there are many legal LLMs [large language models] on the market that seem capable of everything ¨C drafting contracts, reviewing agreements, translating, correcting errors, providing legal opinions, case searches, which all sound impressive. I don¡¯t think that¡¯s right, because if none of these tasks can be done on a level where clients are willing to pay for the experience, you shouldn¡¯t call it a product.

Contrary to common concerns, people are actually very receptive to AI products. The industry¡¯s current lament about low AI penetration primarily stems from products not being user-friendly enough. Products like ChatGPT hardly need promotion; word of mouth is sufficient.

The focus of our industry now should be on gradually enhancing user experiences in specific scenarios, rather than blindly claiming a product can do everything. Technology has not yet evolved to the point where we have one dominant company that can do it all.

As service providers, when clients inquire about a product¡¯s capabilities, our initial reaction may be to say ¡°yes¡±, but it¡¯s equally important to candidly admit ¡°no¡±. This honesty is crucial for a company¡¯s profitability and sustained growth.

Taking Metaso.cn as an example, experience optimisation that we implemented includes: (1) the speed of presenting search results ¨C there¡¯s a significant experiential difference between three seconds and two seconds; (2) the depth of natural language understanding ¨C can the search engine correctly interpret user inputs, even with errors, before providing results?; and (3) the format of search result presentation ¨C does it include unique data sources? For instance, in Metaso.cn, we¡¯ve designed an ¡°academic¡± search option and can also search for related podcast episodes. Optimising these details is based on our continuous observation and response to user feedback.

In addition, experience optimisation efforts must be swift. No company has
absolute technological superiority in the AI services space, so once a company
introduces a new feature, others tend to follow within a month. When it comes to innovation, MetaSota has always been at the forefront, so other companies will imitate us. We have nothing to complain about because that¡¯s the way the competitive environment is.

CBLJ: From your observations, how willing and aware are law firms and in-house counsel regarding legaltech products/services? What are the differences in their needs?

Wang: People have actually received quite a lot of education by now. Today¡¯s discussions no longer revolve around whether AI can replace lawyers, as this issue has been settled, or further discussions would be futile ¨C with technology continually advancing, we will gradually witness different perceptions of this matter.

At the moment, we are all at a stage where we are cautiously assessing the potential for practical applications of AI. I recommend that you try it out for yourself; you can¡¯t judge it by the glamour of its display samples. A good AI product reveals its uniqueness after three days or three consecutive uses.

For example, to compare the difference in quality between our translation service and DeepL and ChatGPT when dealing with different types of documents, one only has to compare three documents and the difference can be clearly perceived. Legal practitioners with specialised knowledge are qualified to quickly see the difference between products.

The needs of clients are actually all-encompassing and multifaceted. A number of our competitors often launch so-called ¡°AI paralegals¡±, which are designed to meet the comprehensive needs of law firms, including the organisation of documents, formatting adjustments, translation.

However, there is no single system that can fully meet all the tasks. If each task can only achieve a level of 30/100 points, it ultimately sums up to zero in user experience; what matters is achieving a high standard of 90/100 points or above in each individual task so as to secure users’ overall acceptance of the product. Will the product sell? It¡¯s easy to answer this question. Invite 30 of your law firm partner friends to use the product and ask them if they would be willing to pay for it, and how much they would be willing to pay. This will give you a good idea of how well the product is being received by the market.

If not even 10 of those 30 friends are willing to pay for it, or if the fees paid are nowhere near enough to cover the cost, then the product is probably not fit for purpose and should not be launched into the market.

In different scenarios, the needs of in-house counsel and lawyers will vary. In terms of translation, their needs are essentially the same, however differences will emerge in more complex tasks.

For example, our MetaLaw¡¯s case search feature is popular among lawyers but has lower demand and average sales among in-house counsel. The primary need of in-house counsel is contract review, a demand that seems less pronounced among lawyers.