Amazon Internet Services bets big on AI
Bikram Bedi, Managing Director, Amazaon Internet Services Pvt. Ltd talks on AI in the cloud
Digit had a chance to sit down with the Managing Director of Amazon Internet Services Pvt. Ltd. and talk about AI in the cloud.
AWS re:Invent 2016 saw a flurry of new announcements in terms new products and upgrades to the existing portfolio, which of them would you put as the most significant from customer interest, specifically in the Indian context?
Bedi: At re:Invent we announced a lot of new services. The ones that are really important for us and are generating a lot of interest from customers would be the services and products that we launched in the Artificial Intelligence space. If you look at AI – Automatic speech recognition, natural language understanding, computer aided vision, image recognition – all of them need some complex machine learning algorithms and large amount of compute capabilities to crunch all the data. So, as an individual developer, one would need to dedicate lot of time and resources to develop this ground up. With our AI initiatives, we are giving developers and startups scale-able solutions around AI along with security, so that developers can focus on creating core applications and their business models instead of worrying about developing AI capabilities from ground-up. Another advantage with our AI services is the pricing. The investment needed to use AI services is directly linked to the scale and does not require huge sums of development expenses. AI is no longer about a few people with lots of money; now even an individual dev can harness the power of AI to make smarter applications.
Indian customers are positive about these new AI based offerings?
Bedi: Yes, absolutely. We are seeing a lot of interests from banks, travel companies and of course internet based startups. Anybody who is looking to interact with their consumers in a more meaningful way would be interested in our AI offerings. For example, with Polly – our text to speech conversion engine – you don’t need customer representatives to answer calls, you can simply use the engine to convert text responses to speech.
Amazon LEX is one of the AI services that offers speech recognition. But there are variations in terms of accents and pronunciations that differ from country to country. So, are you customizing voice engines to understand local languages?
Bedi: Absolutely. We understand the nuances of working on the voice recognition space. For example, Polly supports 24 languages and 47 voices, which includes Indian English. We are continuously localizing our AI engines. We have also been doing a lot of AI and machine learning, at Amazon.com.
Lightsail is another brand new product that was announced at re: Invent 2016. It targets a completely new segment that Amazon Internet Services is entering. Can you tell us more about it?
Bedi: Lightsail has particularly seen a lot of excitement in India. If you are building a basic virtual private server to run some specific experiments and tests, what you need is a solution that starts low in terms of cost but with an ability to scale up if the need arises. Lightsail starts at 5$ a server, it’s a low-end offering and has seen a lot of interest in the Indian market. There are quite a few players in the market with similar offerings, but Lightsail offers a very distinct advantage: it offers inter-connectivity with other AWS services. So, with Lightsail you can start small while having the ability to scale to larger AWS environments without having to start afresh.
When it comes to building new services and offerings, is there a roadmap that Amazon Internet Services follows?
Bedi: We don’t have a particular road-map that we get locked into. Unlike many traditional tech companies, that sort of get bound by the roadmaps, we approach the development of new services and features on basis of our interactions with customers. We constantly talk to our customers for feedback and 90% of our new features and services are based on what our customers are telling us. And we prioritize our product development cycle based on the feedback. For example, we run beta programmes with our customers, giving them access to latest tech and features and in return we get valuable feedback that goes into improvising and developing a strong final product. At Amazon, we firmly believe that launching a service is not the finish line. In fact, it’s just the starting point, we engage with customers for feedback and recommendations to constantly improve the product.
Soham Raninga
Soham Raninga is the Chief Editor for Digit.in. A proponent of performance > features. Soham's tryst with tech started way back in Dec 1997, when he almost destroyed his computer, trying to make the Quake II demo run at >30FPS View Full Profile