Sentient.io launches pay-as-you-go AI services
Singapore-based startup Sentient.io aims to bring the majority of software developers into the artificial intelligence (AI) fold with AI-as-a-service. Its AI and Data Platform, "built by software developers for software developers", allows developers to pick from a growing number of pretrained AI microservices to create smart applications.
This allows businesses to build AI-enabled apps quickly, without the need to train, model and code AI algorithms from scratch. Potential applications include real-time speech transcription in a Singapore-accented English, automatic English subtitling and voice-enabled chatbots which can search a knowledge base using natural language processing.
Christopher Yeo, founder and CEO, Sentient.io and a serial entrepreneur called AI a leveller of capabilities and wealth. He said, "We have a goal of benefiting 1 billion people with the use of AI. We beleive that if we can use our AI-as-a-service platform to reach out to 1 million developers, then they amplify it by each reaching out to another 1 million users of their apps."
"We feel that building AI applications should be easy," he noted. "We want to address that."
Among the challenges of deploying AI are:
- A lack of AI skills
"It takes many years to understand and learn how to do AI," Yeo said. "It's easy to use, very hard to build."
- Software privacy and security
- Integration and interoperability
- Challenges with customisation
Sentient's platform simplifies the way AI is adopted and allows existing AI services from other vendors to be containerised and plugged into the platform, Yeo said. A full catalogue of all AI services can then be displayed and deployed in one place.
The Automatic Speech Recognition (ASR) microservice is Sentient's most popular microservice. It is packaged with features such as transcribed text displayed together with corresponding video and an edit function. According to Sentient, accuracy is above 90%. The technology is trained using the Singapore Infocomm Media Development Authority (IMDA) National Speech Corpus as a dataset. The corpus includes over 2,000 hours of locally-accented audio, 40,000 unique words and local terms. ASR continues to be updated with new additions to the corpus, so it will continue to add value over time.
The Page Profiler microservice is a knowledge base algorithm. It reads news articles and then summarises them for the user. A Document Search microservice can analyse text and identify people, companies and locations.
Sentient.io has worked with A*ccelerate, the commercialisation arm of A*STAR to create a number of AI technologies. It counts A*ccelerate as an early investor and technology contributor. A*STAR is a government agency that supports research and development in Singapore.
While Sentient's origins are in Singapore, most of its microservices can be used anywhere in the world, while the ASR can be adapted to recognise accents from other countries, right down to small groups of people who speak variants of a major language.
Yeo said that developers will be able to leverage on Sentient's domain-specific APIs in 10 minutes. "All you need to do is embed the AI in your app and you're ready to go," he explained.
Users buy credits from Sentient to consume the microservices. Most API calls of a microservice will cost US$0.02 each. The total cost of adopting Sentient.io technology will depend on how many APIs are used and how often the APIs are called with the as-a-service model, while a software licensing model is available for enterprises which want to take the platform into their data centres. Yeo estimated that building a prototype might cost up to US$200 while a live prototype could cost about US$500 a month. Larger enterprises might spend US$2,000 to US$3,000 a month.
Sentient's customers include Brother, Mediacorp and SMRT. The company has handled large projects in Japan and Singapore and plans to have offices in Japan, then Indonesia in future. There is already an office in India.
This allows businesses to build AI-enabled apps quickly, without the need to train, model and code AI algorithms from scratch. Potential applications include real-time speech transcription in a Singapore-accented English, automatic English subtitling and voice-enabled chatbots which can search a knowledge base using natural language processing.
Christopher Yeo, founder and CEO, Sentient.io and a serial entrepreneur called AI a leveller of capabilities and wealth. He said, "We have a goal of benefiting 1 billion people with the use of AI. We beleive that if we can use our AI-as-a-service platform to reach out to 1 million developers, then they amplify it by each reaching out to another 1 million users of their apps."
"We feel that building AI applications should be easy," he noted. "We want to address that."
Among the challenges of deploying AI are:
- A lack of AI skills
"It takes many years to understand and learn how to do AI," Yeo said. "It's easy to use, very hard to build."
- Software privacy and security
- Integration and interoperability
- Challenges with customisation
Sentient's platform simplifies the way AI is adopted and allows existing AI services from other vendors to be containerised and plugged into the platform, Yeo said. A full catalogue of all AI services can then be displayed and deployed in one place.
The Automatic Speech Recognition (ASR) microservice is Sentient's most popular microservice. It is packaged with features such as transcribed text displayed together with corresponding video and an edit function. According to Sentient, accuracy is above 90%. The technology is trained using the Singapore Infocomm Media Development Authority (IMDA) National Speech Corpus as a dataset. The corpus includes over 2,000 hours of locally-accented audio, 40,000 unique words and local terms. ASR continues to be updated with new additions to the corpus, so it will continue to add value over time.
The Page Profiler microservice is a knowledge base algorithm. It reads news articles and then summarises them for the user. A Document Search microservice can analyse text and identify people, companies and locations.
Sentient.io has worked with A*ccelerate, the commercialisation arm of A*STAR to create a number of AI technologies. It counts A*ccelerate as an early investor and technology contributor. A*STAR is a government agency that supports research and development in Singapore.
While Sentient's origins are in Singapore, most of its microservices can be used anywhere in the world, while the ASR can be adapted to recognise accents from other countries, right down to small groups of people who speak variants of a major language.
Yeo said that developers will be able to leverage on Sentient's domain-specific APIs in 10 minutes. "All you need to do is embed the AI in your app and you're ready to go," he explained.
Users buy credits from Sentient to consume the microservices. Most API calls of a microservice will cost US$0.02 each. The total cost of adopting Sentient.io technology will depend on how many APIs are used and how often the APIs are called with the as-a-service model, while a software licensing model is available for enterprises which want to take the platform into their data centres. Yeo estimated that building a prototype might cost up to US$200 while a live prototype could cost about US$500 a month. Larger enterprises might spend US$2,000 to US$3,000 a month.
Sentient's customers include Brother, Mediacorp and SMRT. The company has handled large projects in Japan and Singapore and plans to have offices in Japan, then Indonesia in future. There is already an office in India.
Comments
Post a Comment