Lessons and experiences from building an AI company in Southeast Asia

This is a contributed article by Dr. Huanbo Luan, co-founder and CEO of 6Estates, one of the most hard-core AI startups in Singapore.

Despite being a tiny red dot on the map, Singapore presents both opportunities and challenges as Southeast Asia’s regional hub for anything technology, including AI. Companies in Singapore are able to access other Southeast Asia countries easily.

At the same time, the market in most of the region is not mature enough, and thus not as receptive to advanced technology adoption, especially compared to China and the US. The differences in culture, languages and political environments play a key part in this complexity. 

It is never too early to build AI in Southeast Asia

Let us compare the markets in Singapore and China, especially in Beijing. Singapore is at the early stage of AI evolution and adaptation, thus a less crowded market. At the same time, you will face a group of more professional and open-minded customers in general.

Beijing, on the other hand, is a more mature, aggressive and competitive market, coupled with a much more abundant pool of talents and capital. However, it is increasingly difficult to build up your unique competitive advantages with high entry barriers.

Thus I would say it is more difficult as we need to do more for our clients, but it is definitely not too early. Just like how most countries in Southeast Asia skipped the PC era and jumped directly into the 4G mobile era – I believe that the same will apply to the AI enterprise technology. 

While SEA is slightly lagging behind in digitalization and automation as compared to the more developed regions, there is no reason for the equally tech-eager big enterprises in this region to wait for all steps to happen one by one before moving into AI.

In fact, it makes perfect sense for them to go directly into the intelligent automation world when it comes to implementation. However, it can be advantageous sometimes to start from zero to build the digital and automation infrastructure dedicated for AI.

The COVID-19 impact

Challenges are at every corner for any startups. The most difficult one is to be able to transform technological advantages into real opportunities, and this Covid-19 global pandemic makes it particularly difficult for us.

The challenges come in all directions. First is the internal team. When the Covid-19 pandemic first started, everyone was unsure what would the impact be and how long it would last. Then it comes to travel restrictions, social distancing, and work from home, etc. – every individual has to adjust his/her work and lifestyle to adapt to the new social norm. It is a challenge to keep the team coherent and motivated.

The second is the external market. Many industries are severely affected by the out-break of Covid-19. Large enterprises are no exception as all plans and budgets need to be reviewed and adjusted, and it means delays and uncertainties to the sales process.

Moreover, the capital market is volatile and early stage venture capital is almost completely shut down for the initial few months since the out-break of Covid-19. One silver lining of this is forcing everyone in this region to think seriously about their digitalization strategy, which we believe, will bring in more opportunities in the mid-term of this global pandemic.

Other than that, finding the best talents in this region is as difficult as anywhere else in the world. High quality AI talents are scarce resources, especially the mid-level talents. Fortunately, there are four world-renowned universities in Singapore, providing a decent pool of talent that we can tap upon.

A skillful team backed by AI/NLP experts 

As a company rooted in Tsinghua-NUS joint AI research centre, we have vast access to China’s talent market, allowing us to set up R&D teams in both Singapore and China. Most of the co-founders and senior team members are profound holders of PhD degrees in the AI technology discipline. The team has accumulated over 10 years of experience in AI/NLP research before founding 6Estates.

Our team remains active in frontier research. For instance, we have contributed multiple AI papers in the past several years to top AI/NLP conferences. This has been our unique competitive advantage in both business development and talent attraction.

Our engineers know that they will be supervised by top AI scientists and they will always be working on the most frontier technology in 6Estates. This helps us to recruit, develop, and retain top AI engineers.

Making magic for Financial Institutions

AI, as we know it today, is applicable to many sectors and many business activities. A lot of effort has been spent on adapting our technology to solving real world problems – only then AI can be valuable to the clients as well as the society at large.  

For that, we have been focusing on Machine Learning and NLP as well as their applications in FinTech. 

Our current product suite provides intelligent process automation in financial applications. In particular, our product is especially focused on reading and processing complex business documents, and converting unstructured data to structured data.

At this stage we have two market-ready products. First one is “LC Automize” that targets trade finance businesses. It helps the trade finance operation teams in banks and other financial institutions improve efficiency and reduce human error. It processes scanned supporting documents for Letter of Credit (LC), automatically extracts the target fields, and performs rules and compliance checks.

The second product is “FA Automize” which is tailored specifically for financial analysis related tasks. It takes financial documents as input, such as annual and quarterly reports. It then performs financial spreading to extract key financial numbers and reads through the documents to perform financial analysis. Finally, it will auto-generate draft summary reports according to the user’s requirements. 

The pipeline and the resilience

We have been receiving requests from various industry sectors, hoping that we could address their current challenges and solve their problems. With a solid technological foundation and a team of very experienced senior AI leaders in the industry, we are confident to gradually bring more products to the market.

The initial phase where we developed the first commercialised use case took a long time – and for the right reasons. We technologists tend to approach the world differently from real business decision makers – and aligning the priorities of both parties takes time and effort. We have gone from extreme confidence to multiple levels of emotional downturns over the years; and I am glad that we are now on the right track to deliver more value to our clients in multiple sectors. 

Based on our experience, our advice to other aspiring genuine AI companies would be “to be patient and get ready for the magic”. This is because adopting AI solutions is exactly like taking in new hires – you can start to enjoy a highly efficient performing team only if you have invested adequate time to nurture them first.

If you read Andreessen Horowitz’s book “The Hard Thing about Hard Things” – you would know that the most important quality of any great CEO is resilience (not calling quit). This especially applies to great AI startups especially – as often it takes years for you to get the first product ready, but once you are there, you will enjoy a pretty steep (and exhilarating) take-off. 

If you have problems which you believe our technology might be of help, or use cases that we can develop together, please contact me and I will be happy to explore together with you.