Source: SenseAI
Image Source: Generated by Wujie AI
"TheB.AI" provides an integrated artificial intelligence platform that supports various AI models, including advanced language and image models. Users can integrate multiple large models into their workflows through web applications or APIs. The platform offers real-time search, image generation, customizable model personalities, and long-term memory functions. It has both free and paid features, with costs calculated based on actual usage. In June 2023, TheB.AI entered the global Top 50 GenAI web product released by a16z, ranking in the Top 20.
This issue features an exclusive interview with TheB.AI founder Wang Gang, exploring technical considerations and applications together with Sense AI.
01. Founder Background Introduction
Sense AI: Please briefly introduce yourself.
Wang Gang: Hello, everyone. I'm Wang Gang, born in 2004. Currently, I mainly engage in entrepreneurial work. My entrepreneurial story began at the age of 14 when I opened a store on Taobao, specializing in selling Xbox game CD-Keys from Argentina and Turkey. During that time, I not only developed my business sense but also developed a strong interest in technology. Subsequently, I noticed that many Chinese people wanted to register overseas companies, so I founded a company specializing in providing this service. After that, I ventured into the field of virtual credit cards, helping businesses complete international transactions. Last year, inspired by the release of ChatGPT, I started my latest project—TheB.AI. This is an artificial intelligence platform that integrates GPT-4 and other advanced models. This project stems from my passion for AI and programming, aiming to provide users with a convenient, integrated AI experience. TheB.AI not only supports multi-model interaction but also introduces innovative features such as long-term memory and plugin systems. In the field of AI, I have been exploring more possibilities.
Sense AI: Entering the market for entrepreneurship at an early age, what experiences and gains do you have?
Wang Gang: The project that brought me the most gains was my first project on Taobao. Through this project, I learned how to identify and seize opportunities in a competitive market. I was one of the earliest merchants to import Xbox game CD-Keys from Argentina and Turkey, and this experience taught me how to understand market demand and trends. Additionally, this project sparked my interest in technology, prompting me to start learning programming on my own. This laid a solid foundation for my subsequent establishment of TheB.AI. Apart from my Taobao project, my experiences in helping people register overseas companies and VCC brought me significant gains. This phase taught me how to identify and meet the specific needs of the market, especially in the field of cross-border commerce and financial services. Through these experiences, I learned how to handle complex legal and financial issues, which greatly helped me in dealing with technological and business challenges when creating TheB.AI. Furthermore, each of my projects emphasized the importance of user experience, and I have always been committed to simplifying users' operational processes and improving their satisfaction. This user-centered mindset played a crucial role in the development of TheB.AI, helping me design an efficient and user-friendly AI platform.
Sense AI: How did you first become aware of the opportunity with large models, and how do you view the potential and timing of this technological wave?
Wang Gang: My initial interest in large models was mainly due to my long-standing interest in artificial intelligence and machine learning. Models like BERT and GPT-3 not only excel in understanding and generating language but also demonstrate remarkable capabilities in integrating knowledge, creating content, and solving complex problems. I remember the first time I used ChatGPT, I was attracted to its depth of language understanding and natural fluency in answering questions. It not only understands our intentions but also provides appropriate responses. I believe that this kind of technology will soon become a part of our daily lives, whether it's handling documents or writing code, it can help us improve efficiency. As for the timing of entrepreneurship, I believe that the earlier one gets involved in chatbots, the better. Early involvement means being able to enjoy the dividends of natural traffic, and the cost of acquiring customers is also very low. As Silicon Valley has a saying: "Act as fast as possible."
02. Project Introduction
Sense AI: When did you become aware of the opportunity with large models, and how did you choose your entry point?
Wang Gang: My interest in large language models began with Meta's release of Galactica. Its capabilities in data processing and knowledge integration deeply attracted me. Although Galactica's final results were not as expected, I quickly saw its enormous potential in practical applications. I chose to enter this field by combining my market insights and personal interests. When ChatGPT was launched, despite its unstable service and limited functionality, it precisely showed me the opportunity to provide an easy-to-use, feature-rich AI platform. Therefore, I created TheB.AI. We not only allow users to access multiple advanced models but also introduce innovative features on the platform, such as long-term memory, multi-model switching, and plugin systems, to meet specific market needs. Through continuous learning and market observation, I found my entry point and built TheB.AI on this basis.
Sense AI: There are many products for model platforms and tool platforms. What is the biggest differentiation of TheB.AI, and I'm also curious about the origin of this name?
Wang Gang: TheB.AI's unique feature lies in how we integrate different AI models and allow users to personalize these models according to their needs. When AI was just emerging, many companies targeted the entertainment industry. However, I personally believe that while AI entertainment is attractive, it may be short-lived. I am more optimistic about the potential of AI in improving work efficiency. Therefore, TheB.AI is committed to integrating cutting-edge artificial intelligence technologies like GPT-4 and providing collaborative functions, such as a powerful plugin system. Users can experience the capabilities of various AI models on our platform and customize them to meet their specific needs, whether it's engaging in coherent deep conversations or performing specific tasks, such as writing code, handling documents, or integrating Wolfram Alpha's scientific computing capabilities. As for the name "TheB.AI," it reflects my vision. In my view, "B" represents "Beyond" and also implies "Best." This name is not just a label but a commitment to our pursuit of technological limits and innovation. TheB.AI symbolizes our firm determination to surpass the traditional limitations of artificial intelligence, in short, striving to achieve "Beyond the Best Artificial Intelligence."
Sense AI: How do you view platforms like Replicate and Open Art, or have you found similar products that you think are doing well?
Wang Gang: Regarding platforms like Replicate and Open Art, I believe each has its own strengths. Replicate is mainly targeted at developers, providing a unified interface for the convenient deployment and hosting of various models, whether for image generation or natural language processing. This service model facilitates innovative experiments for developers, and the usage-based billing system is more flexible. However, there is room for improvement in the user interface's friendliness, and its service has a high degree of substitutability.
On the other hand, Open Art may have broader development opportunities in the design and entertainment industries, but it also faces challenges in terms of profit models and long-term commercialization.
In my view, products with long-term development potential, such as Together.AI and Modal, are worth more attention. In an evolving future, models will always need reliable platforms to support their operation. I continue to maintain close communication with users and have realized that the development of Artificial General Intelligence (AGI) will not happen overnight but will gradually strengthen and eventually integrate into human society. Products developed with this progressive perspective and new thinking, I expect, may be able to meet everyone in the near future, possibly within six months.
Sense AI: What were the earliest users like, what are their common needs, and what is the current ratio of C-end and B-end users?
Wang Gang: In the initial stage, TheB.AI attracted users who were early adopters familiar with ChatGPT and were interested in exploring similar technologies, often trying our platform for entertainment purposes. Over time, these users evolved into professionals seeking to improve efficiency in their work, such as using our services for document writing (such as daily and weekly reports and text content optimization), programming (helping interpret errors, code adjustments, and organization), and educational learning (language learning and summarizing and answering academic papers). Currently, our customer base is mainly composed of C-end (individual consumers) but also includes some startups and research groups who integrate our API or use our web application to support their teams or departments. While individual users make up the majority, the proportion of B-end users (business customers) is steadily increasing.
Sense AI: There are actually many tools that meet these needs, and there are vertical tools as well. How does TheB.AI find a balance between performance experience and cost if it wants to do better?
Wang Gang: To find the best balance between performance, user experience, and cost, TheB.AI is continuously committed to innovation and technological optimization. We provide users with customized model configuration options to meet the specific needs of different user groups through modular functionality. This customization not only increases the platform's flexibility but also ensures that users only pay for the features they need, effectively controlling costs.
We use a flexible billing model based on users' actual resource consumption, providing affordable entry-level services for startups and meeting the high-concurrency demands of large enterprises. Technologically, we have deployed an intelligent caching mechanism to shorten response times, ensuring that users can enjoy a fast and smooth operational experience. Additionally, we have introduced a plugin system to maximize productivity and improve work efficiency. In the platform's construction process, we have combined in-house development with open-source community resources, reducing development costs and ensuring that our technology stays at the industry forefront.
Most importantly, we value user feedback. We maintain close contact with users, actively listen to their opinions and suggestions, and ensure that our services are continuously optimized around their actual needs. This user-friendly, open attitude is key to finding the balance point.
Sense AI: How do you view OpenAI's GPTs and API ecosystem, which also has a large number of tools and is built on such a large traffic entry?
Wang Gang: Firstly, traffic is indeed a significant advantage as it provides extensive exposure and usage scenarios for OpenAI's technology and services. This wide application may promote rapid iteration and optimization of the technology and also bring feedback from users in different fields, which is crucial for the maturity and development of any technology. However, traffic is not a decisive factor. The maturity, reliability, and user experience of the technology are equally important. For OpenAI's GPTs, their low entry barriers allow more people to easily access and use the technology, but this may also lead to a large number of repetitive and low-quality applications in the market. This phenomenon is common in any rapidly developing and popular technology field, and applications of varying quality may affect users' overall perception of the value of GPTs. I believe that OpenAI should actually focus more on the development of next-generation models.
Sense AI: How can you provide a personalized experience and incorporate domain knowledge without directly fine-tuning the underlying models?
Wang Gang: We can use "Retrieval-Augmented Generation" (RAG) technology. RAG technology enables large language models (LLMs) to retrieve the most accurate and up-to-date information from external knowledge bases without changing the original architecture, and it has shown powerful effectiveness in integrating domain-specific knowledge. As for personalized experience, this requires a deep understanding of the unique needs of different user groups and exploration of the specific aspects in which they expect personalization.
Sense AI: From our current perspective, what are the characteristics and strengths of different foundation models?
Wang Gang: In my usage and user research, GPT-3.5-Turbo is suitable for most lightweight tasks, role-playing, and casual conversations, among others. However, when we pose more complex and unique questions, we often need a model of the level of GPT-4 to get better answers. GPT-4 is much better in terms of logic and mathematical abilities than GPT-3.5-Turbo, but it has shortcomings in text composition. Those who frequently use GPT-4 may know that it tends to "overdo" the text composition, and Claude performs much better in this regard.
Sense AI: Will you consider integrating more models into the platform in the future, and how do you prioritize the integration of models? Will video generation be included in the future?
Wang Gang: Absolutely. We are committed to continuously enriching and optimizing our platform to meet the needs and expectations of our users. When introducing new models, we mainly follow the following principles:
- Universality - First, we prioritize models that have a wide range of applications for users.
- Domain expertise - Secondly, we look for models that demonstrate significant advantages in specific professional fields such as programming, which should provide deeper capabilities than existing models like GPT-4.
- Inheritance/extension capabilities - We also value the integration capabilities of models, especially their potential for integration with external tools, as this is crucial for extending the practicality of the models.
Regarding video content generation, we are monitoring and researching text-to-video and image-to-video conversion technologies. We plan to integrate video generation into our system when these technologies mature enough to provide stable, high-quality output. We look forward to being able to create detailed and accurate video content based on concise text descriptions in the near future.
Sense AI: What are the most important and challenging engineering difficulties in our development?
Wang Gang: In our development process, the most challenging part is not the technical implementation itself but the innovative thinking behind it. As early as March last year, I began to build a long-term memory system and a plugin system, even before ChatGPT launched their official plugin store. As an independent developer, the main challenge I face is the breadth and depth of code writing; sometimes, the workload is quite substantial, haha.
Sense AI: How does TheB.AI commercialize, and is it profitable?
Wang Gang: TheB.AI's commercialization strategy is not only focused on achieving profitability but also on sustainable development and deep optimization of services. Our business model is primarily designed around user usage, using a Pay-As-You-Go model. The benefit of this model lies in its flexibility and fairness—users only pay for the resources and services they actually use. This not only lowers the barrier for new users but also allows large users to conveniently expand their usage when necessary. TheB.AI has achieved a balance between revenue and expenses and has already achieved a small profit. We believe that profitability is not a short-term performance indicator but a direct reflection of long-term value and customer satisfaction.
03. Product Growth Strategy
Sense AI: How did TheB.AI achieve rapid growth in the early stages? What are the natural growth factors, and was there any paid advertising?
Wang Gang: Our focus has been on organic growth, and we have not engaged in any paid promotion. The key to TheB.AI's significant growth in the initial stage lies in our timeliness—we promptly launched free and stable services. These services, driven by user feedback and strong community discussion, have led to natural user growth. User discussions on technical blogs, professional forums, and social media have provided us with significant exposure and have also helped build TheB.AI's brand reputation. This reputation is built on genuine user satisfaction and recognition of the stability and practicality of our services.
Sense AI: What is the current scale of TheB.AI, and how confident are you in achieving a larger scale in the future?
Wang Gang: TheB.AI currently has millions of monthly active users and is experiencing continuous monthly growth. As for how large we can expand, I hold a cautiously optimistic attitude. There has not yet been a disruptive "killer app" in the industry, but I firmly believe that the key lies in continuing to deepen and optimize our existing productivity tools. By deepening our focus on work scenarios, taking steady steps forward, closely monitoring market dynamics, and continuously respecting and driving the development of technology, we are following our current strategy.
04. Divergent Discussion
Sense AI: In what direction might TheB.AI expand in the future, and if AGI were to appear one day, would our positioning change?
Wang Gang: As I mentioned earlier, TheB.AI will continue to steadily focus on deepening work scenarios and accelerating productivity. If AGI were to appear, what all AI startups are doing now would become meaningless. As I mentioned, I believe that AGI is a gradual evolutionary process and is unlikely to be a one-time technological leap. In this gradual transition, we can continuously evaluate and adjust our strategy.
Sense AI: How do you view the relationship between the closed-source ecosystem of major companies and the open-source community ecosystem in the future?
Wang Gang: Regarding the relationship between the closed-source AI ecosystem of major companies and the open-source AI community ecosystem, I believe that these two are not a zero-sum game but can complement each other. The closed-source AI ecosystem, especially products launched by large companies, often represents the technological forefront of the industry and provides highly integrated and specialized solutions. These products usually have stronger stability and performance, suitable for user groups with high requirements for AI accuracy and reliability.
On the other hand, the open-source AI community ecosystem focuses more on innovation and flexibility. It encourages global developers to share knowledge, collaborate on problem-solving, and quickly adapt to market changes. Open-source projects can more easily obtain feedback and improvements from the community, which is crucial for driving the rapid development of AI technology. Additionally, the open-source AI ecosystem also provides possibilities for individuals or small teams with limited resources who want to quickly experiment with new ideas.
TheB.AI plays a bridging role in this. We are committed to integrating closed-source and open-source AI technologies to provide users with a professional and flexible platform. We believe that through this approach, we can help users find the most suitable solutions for themselves in the complex AI ecosystem, whether it's using cutting-edge closed-source technology or flexible open-source tools.
Sense AI: Are there any AI tools that you personally think are good?
Wang Gang: Among the AI tools I have experienced, I think three are very good: Rewind, Immersive Translation, and MetaGPT.
Rewind: This tool has been very helpful to me as it can track and record my daily activities and automatically generate daily reports. This is particularly important for me when I need to recall details of interesting content I have read but have forgotten after some time. I can quickly obtain past information by simply asking Rewind.
Immersive Translation: This is an outstanding translation plugin that has greatly improved my experience of browsing Russian and Japanese websites. Its high translation accuracy allows me to understand foreign language content more deeply, enhancing my efficiency in browsing foreign language websites.
MetaGPT: As a multi-agent framework, MetaGPT has sparked my great interest. I often use it to help me summarize various reports, and its powerful features make my work more efficient.
Sense AI: Can you share what you think are the three most important elements in AI entrepreneurship?
Wang Gang: When discussing the key elements of AI entrepreneurship, I believe we can focus on the following three points:
First is "swift action." The development pace in the AI field is very rapid, and entrepreneurs need to quickly translate ideas into actual products or services to seize opportunities.
The second element is "deep insight," which means that entrepreneurs should have a broad knowledge base, closely monitor industry trends, and continuously learn and understand user needs from market feedback.
The last but equally important element is "perseverance and patience." On the path of AI entrepreneurship, setbacks and failures are inevitable, but continuous efforts to optimize products and patiently wait for market recognition are key factors for success.
免责声明：本文章仅代表作者个人观点，不代表本平台的立场和观点。本文章仅供信息分享，不构成对任何人的任何投资建议。用户与作者之间的任何争议，与本平台无关。如网页中刊载的文章或图片涉及侵权，请提供相关的权利证明和身份证明发送邮件到support@aicoin.com，本平台相关工作人员将会进行核查。
