When we hear the terms “artificial intelligence” and “natural language processing,” we often think of innovations coming out of the U.S., China and Japan, but not the Philippines—an impression a local technology startup aims to change in the next three to five years.
“The Filipino AI software developers are world-class, and it’s high-time that they are recognized,” said Gian Dela Rama, chief executive officer of InnoVantage, developer of Nudge, a process automation platform wrapped by an AI chatbot. “And I think that’s going to happen within the next three to five years given the pace of development.”
Dela Rama, who had no formal education on information technology, co-founded InnoVantage five years ago with the goal of helping companies turn paper into electronic forms, a process known as digitization. “[Our clients] wanted to go faster, be more efficient, and be more responsive to customers. But they couldn’t really achieve that because everything was routed with a piece of paper,” said Dela Rama in an exclusive interview with UpgradeMag.com.
InnoVantage’s service expanded from digitization to process automation after one of its corporate customers wanted an automated process that did not involve typing information on digital forms. “That became the seed that bore Nudge. We re-developed the platform to automate business processes using natural language processing. So instead of filling out a form, you chat with an AI bot. And that bot will be the one to do the routing, the working, the notification, reporting and all that.”
While there are many off-the-shelf AI chatbots in the market today, InnoVantage wanted to develop its own AI platform. “It took us about four years to build the platform, which is too long for a startup. Normally, you can come up with a viable product within a few months. Yes, we had that option but we wanted to have our own intellectual property. Since we are serving large enterprises, we needed to own the code so we could be more flexible in meeting the requirements of clients, especially their data security and privacy concerns.”
Now that the company is servicing more companies, the investment has paid off in both the flexibility of the platform as well as the scalability which was mostly provided by Amazon Web Services, according to Dela Rama who has a degree in business management. “We are already branching out and we couldn’t do that if we weren’t on the cloud.”
Today, Nudge is live in 15 countries within Southeast Asia (SEA) and Europe, the Middle East and Africa (EMEA). The platform, developed using Java and Python, is used by companies in regulated industries, such as financial services, banking, pharmaceuticals, insurance, healthcare, telecommunications, and government.
Currently, InnoVantage is also working on expanding Nudge’s language platform to include Mandarin and local dialects such as Ilocano, Cebuano, Kapampangan, to name a few.
“AI is fairly new and our bots, admittedly, still sound ‘boty’ but it’s getting more and more human-like; and a lot of that is actually because of the research we are doing on natural language processing,” said Dela Rama.
Born in the cloud
Like most technology startups today, InnoVantage was born on the cloud. “All of our software and tools came from the public cloud. Nudge was born on AWS,” said Dela Rama. “AWS was our first cloud provider, but we went to Azure as well because our first client was on Azure. But when we got our software to the level of technology we were comfortable with, we felt that AWS was a better fit because of its capabilities. So right now our production is 100% AWS.”
Of AWS’ more than 90 cloud computing services, InnoVantage uses around 20 of them, namely: Amazon Elastic Compute Cloud, Auto Scaling, Elastic Load Balancing, Amazon EC2 Container Service, Amazon Simple Storage Service, Amazon Elastic Block Store, Amazon RDS, Amazon Dynamo DB, Amazon API Gateway, AWS Identity & Access Management, Amazon WAF, Amazon Shield, Amazon Machine Learning, Amazon Route 53, Amazon CloudFront, Amazon Simple Email Service, Amazon Simple Notification Service, Amazon WorkSpaces, Amazon CloudWatch, and Amazon CloudFormation.
The combination of most of these services have allowed increasingly seamless interactions, prompting only 3-second response times for the Nudge chatbots and efficient content delivery and analysis during real-time customer conversations.
“A benefit to us, which is not readily apparent to the average person, is the speed in which we can change our infrastructure,” said Dela Rama. “Because normally, if you were doing it via physical servers, once you buy you have to work within the constraints of the box. If a client shifts direction it would be expensive to shut down the servers. Because of AWS, we are able to move with our clients. The [Amazon EC2] allows us to breathe in the sense that it lets us expand and contract.”
Dela Rama also said that with Amazon Elastic Block Store (EBS) and Amazon S3, they no longer have to care about storage the normal way. “To be honest, right now we don’t care. Because whatever storage requirement we have, we know it’s going to fit.”
As for the database, Dela Rama related that the biggest advantage of either Amazon RDS and Amazon DynamoDB is the absence of capex. “We don’t have to buy an actual database. When we started the company we had very limited resources. We couldn’t buy a physical server even if we wanted to because we didn’t have the money to do that. We know that the server, even if it’s expensive, is not the largest cost within the application, it’s the actual application within the database that might cost more.”
While Amazon EC2 can run machine learning software it wasn’t designed for machine learning, which prompted InnoVantage to migrate its training platform to Amazon Machine Learning, said Dela Rama.
“Amazon Machine Learning allows us to train the neural network models that we built in a platform that gives us the flexibility of an EC2 and the raw computational power required to do model training. Previous to that we were using the GPU instance of AWS which is probably their highest-performing class of EC2 servers. It’s good but it’s not as special purpose as Amazon Machine Learning.”
For business continuity purposes, Nudge has been designed to be in multiple Availability Zones of AWS, according to Dela Rama. “A lot of the components of Nudge have to be up and running to be able to give a response in 3 seconds. That’s the reason why we rely on the resiliency or uptime of the AWS servers.”
While Dela Rama is very satisfied with AWS services, he would like the company to improve their documentation’s search facility. “Also, now that they have more than 90 services and growing, it is getting more difficult for us developers to keep track of what is happening. I think they should have a service that would train developers on how to use AWS. We need to maximize the service we are using.”
An AI platform that does more than just chat, Nudge has the potential to reach more companies in countries beyond SEA and EMEA, and eventually help put the Philippines on the world’s list of top AI developers, as what Dela Rama envisions.