How to win deals from Banks for your Regtech Startup
Find business opportunities for them
According to KPMG, only 20% of the population in Cambodia have a bank account.
(https://theaseanpost.com/article/banking-southeast-asias-unbanked). Despite this astounding statistic, Cambodians are quite tech savvy. Data collected in September 2016 by USAID and partners show that over 96% of the Cambodians claim to own their own phone, and more than 99% are reachable through some sort of phone. Moreover, at least 50% of the Cambodians claim to have accessed Internet or Facebook. (http://www.open.org.kh/research/phones_2016.pdf ).
This presents an untapped business opportunity not to be missed by any institution that offers some sort of finance and banking service. Mobile banking is the future of SouthEast Asia. Additionally, mobile banking that can be done in social media platforms like Facebook have even more far reaching potential for the future.
ABA Bank Cambodia, a bank backed by National Bank of Canada, was one of the first banks to realize this future market. An idea of developing an AI powered auto bank application via Facebook Messenger came into fruition when they approached Growthbotics, an AI solutions company for finance, to discuss about customized solutions.
ABA Bank wanted an AI powered conversational chat assistant that lives in Facebook Messenger. The chat assistant would be able to with general banking inquiries as well as help new customers apply for bank accounts. Applying for a bank account online is no easy task. Banks biggest pain point is complying with international banking law that includes Know Your Customer (KYC) and AML (Anti-Money Laundering). With Artificial Intelligence, KYC and AML would be easier for the bank. Not only did we need to build out the Natural Learning Understanding (NLU) for the AI, but we had to build out the entire banking application ONLINE. This required the development of several in-house APIs, AI image/face recognition and live chat CRM support, in case the chatbot does not understand customer inquiries. We also needed to develop secure databases to store IDs, passports, customer information.
Get on the same page, with flow charts and diagrams
First things first, laying out a systematic method of communication with the client is very important. Growthbotics’ had to build a demo chatbot using Chatfuel while ABA Bank used a spreadsheet to list out the conversation flow. Growthbotics also drew out the chat logic flow. Below is part of the flow for choosing English or Cambodian as the preferred language:
Being on the same terms with the client is harder than it sounds. It requires a lot of back and forth communication with the client which involves a lot of skype calls and meetings. In our experience, at least 8 iterations are required until a client closes a project. Below is a snippet of the draft chat flow in Word document format that had been modified for at least 8 times:
After seeing everything eye to eye with ABA Bank, Growthbotics started hammering out the first prototype. The milestones were based on the features they would deliver. The first feature was Image Recognition for ABA Bank’s customers. In order to open a bank account, customers need to upload their passport or Cambodian national ID. We also needed to develop an API that would cross check the uploaded IDs to see if any IDs are blacklisted by the bank:
Be good at what you say you can do
In addition to image recognition, we also developed facial recognition to cross check the ID with the face. Facial recognition also requires AI technology to detect human faces and to interpolate photo IDs with selfies and face photos:
In the next blog, we will continue with this AI chat assistant for ABA Bank and specifically explore the native APIs that Growthbotics had built for their client.
Like our blog post? Send a message over on growthbotics!
