About AI Monetization: The Market, Customer Needs, and How To Close Deals

Last Sunday, EqualOcean organized the WIM Salon x Beijing to bring together industry players to talk about monetizing AI and other key stories affecting technology. The Salon brought together innovators from AI companies Terminus, Transwarp and Shannon.AI for a productive event.

Terminus is a technology company, focusing on combining AI and AIoT technology with traditional industries. Mr. Lex Xie (谢超), VP of Terminus, has rich experience in IoT platforms. Before Terminus he worked in Accenture, and also had a ten-year successful Internet entrepreneurship career and nine years of experience in mobile Internet business operation and management.

Transwarp is a big data company with a focus on cloud computing and AI. The company currently covers many industries, leveraging computer vision, IoT technology, traditional machine learning, etc. Dr. Yang Yifan (杨一帆), Chief Product Officer of Transwarp, is experienced in the statistical learning, deep learning and graph computing fields and has published more than ten papers (SCI and top conferences). Before joining Transwarp, he worked in Bank of America and Alibaba.

Shannon.AI started at the end of 2017, focusing on NLP in the finance industry. To structured Shannon.AI uses end-to-end deep learning to process data and provide services to financial institutions. Ms. Sophie Bu (⼘贺纯), business partner of Shannon.AI, received her MBA degree from Stanford Graduate School of Business in 2017. She led sales functions in many areas, including developing Belt and Road projects based in Kenya and Ethiopia. She joined the company in 2018.

Andrés Rodríguez, co-director of Startup Grind Beijing, was the host for the discussion.

Key Takeaways:

Market trend: Do we have too much data or too little?  

Data shortage: Data is the problem of making further progress. People want to know how much to pay for data, how to collect data; that’s why we take advantage of IoT. There’s a data loop in AIoT: from transformation to online, intellectualization to automation. Data can provide insights as illustrated in three scenarios:

First is to do with caring for vulnerable and senior citizens in a residential area. Traditionally, volunteers visit senior people once a week or twice a week, but it’s hard to detect an emergency. Traditionally, volunteers ask senior people to put fresh flowers on the balcony and to know they are doing well. Now by monitoring the entire residential area, if senior people don’t show up in 24 or 48 hours, volunteers are alerted and can check them immediately. 

Second, considering the financial industry, what if we have more information when measuring risk for small to medium companies? What if we can know more about the visitors to the company, the storage of inventories, the usage of electric devices, the number of employees? 

Third, fire control: when a fire alarm happens, we know which walls are breakable and where the closest fire hole is. Because of IoT data, we are able to put the whole scenes together. (Mr. Lex Xie)

Data integration: There is a strong issue of data isolation. Taking the financial industry as an example, there are multiple sources of data, including images, natural language data, structured or unstructured data. How to integrate them into one standalone data platform? Knowledge graphs help us to connect. We provide the graph database, which is very different from the traditional key-value database. Customers also want algorithms to be integrated. We are required to provide a whole system that deals with machine learning and deep learning. (Dr. Yang Yifan)

Data explosion: we are transitioning from an age of scarcity to an age of abundance in terms of data. How to find all the information from all the different tools at our disposal? NLP lowers the costs and barriers to finding useful information. By using NLP and creating a user-friendly interface, we make it easy to find useful data. All the sophisticated technologies are in the back. In the financial industry, individual investors don’t know how to use databases such as Bloomberg and Wind, nor how to pull MySQL data. We help individual investors and professionals to conveniently get information. We are also working on alternative data, unstructured data that’s hard to access and collect. (Ms. Sophie Bu)

R&D and customer needs: A simple user interface is what they want. From the R&D perspective, algorithms are simple. Models are only 30% of the solution. During the process, we start from searching the market, by talking to risk managers in finance industry; besides integrating isolated data, what they want is a simple tool, a UI tool. Leveraging AI+UI will help us create a lot of products that satisfy customer requirements. Customers want to utilize data, try different algorithms, and use these functions on one platform. Then we test and get feedback, keeping modification and updates, to ensure the result is usable. (Dr. Yang Yifan)

'Making it easy to use' is very important but not sellable. Combining user-friendly interface with data or hardware is precious. (Ms. Sophie Bu) 

Pricing, closing the deal and making money

Pricing: In sales, there's 'death valley' theory: you either sell each deal for millions to big enterprises, or you sell at a very low price to the mass market. 
For large enterprises, there are multiple decision-makers and they have different opinions about the products. So for big companies, the strategy is to sell sophisticated products, such as a platform for machine learning and deep learning tools, because these enterprises tend to have less AI talent but many IT people. There are several things to consider. Do they have a large enough budget? What is the priority? Who makes the decision? Since we are selling something very new, we need to coach our clients and even CTOs. So we need a team to do market research. 

For smaller companies, we sell APIs instead of customized solutions. So the pricing is different. We can benchmark competitive products, always testing the water and charge the price. The most important thing is to not stay in the middle of ‘death valley’. (Ms. Sophie Bu)

Closing clients: The first step is to give a concept. It’s impossible to build a smart city now because it’s too complicated, but we can divide the whole smart city part-by-part and start by making a small part intelligent first, such as a campus, a building or a residence, and later to integrate many of these together. So we first let the client know the concept of building a smart city. 

Second is about figuring out the right group of clients. We want to provide solutions and get money back. We do not provide free services or invest resources over years with no return. It’s very dangerous for startups to burn money. 

Third, companies do not always care about technical problems and economic problems – but decision-makers also care about personal problems. If there is a serious fire, there's not only an economic loss for people, but it’s a loss of someone’s future career. We want to let our customers know their hidden losses for not using the new technology, whether provided by us or not. (Mr. Lex Xie)

Going global: Building trust is important. From the beverage industry to Belt and Road projects to AI companies, it’s always important to build trust with your clients. We need to let them know our products are capable of solving problems. To build trust, there are many intellectual exchanges, from sending the right people to technical discussions – there is a need to build demos as fast as possible. We already have an international team, but there are still physical location barriers for now. (Ms. Sophie Bu)

Inspiration and innovation are two keywords. Ten years ago, there was no deep learning. As we keep improving and remain open, the market will grow bigger and become healthier. (Dr. Yang Yifan)

The IoT market has a bright future as every traditional device and facility can be replaced by IoT devices. But we need to choose the right industry and timing to apply the cost-effective technology to it. (Mr. Lex Xie)