Artificial intelligence (AI) is rapidly transforming everything today, from daily lives to transportation to businesses. Humans have always found the concept of AI very enthralling as is evident from the number of hit sci-fi movies. Scientists and researchers have worked hard on making this technology a norm for the human beings.
Enterprises are adopting AI and machine learning (ML) for various use cases, which has risen the demand for AI engines that can be used to develop intelligent applications and tools. Such apps and tools help them automate the repetitive, tedious and difficult tasks that can affect productivity and cost of operation.
Automating workflows also help enterprises in reducing human error and boost efficiency. Today, everybody is familiar with AI tools like Siri, Cortana, Alexa, and Google Assistant.
Software developers are learning new AI and ML skills to keep up with the pace and grow along with the emerging technologies. To build new AI projects or take existing projects to the next level, there are a wide range of AI engines available for developers.
These engines can be used to build several AI applications, like a personal assistant, chatbot, to connect to smart home, control devices, and more. Today, there are several options available for developers to build their own AI project.
4 best artificial intelligence engines for 2019
1. Microsoft Cognitive Services
Microsoft provides intelligent APIs which developers can use to infuse vision, speech, language and knowledge capabilities into applications, websites, and bots.
Part of the Microsoft AI platform, the Microsoft Cognitive Services, allow developers to compose intelligent applications, that can be customized according to the availability, security and compliance requirements of the organization.
Microsoft Cognitive Services include the following:
- Emotion API
- Computer Vision API
- Language Understanding (LUIS)
- Speaker Recognition API
It detects the expressions on faces in an image and returns the confidence across a set of emotions for each face in the image. It also uses Face API to bound the face with a box.
|Pricing (Emotion API) :|
|Free||30,000 image transactions free per month. 3,000 status queries per month.|
|Basic||10 image transactions per second ($0.10 per 1,000 transactions)|
|Standard||10 image transactions per second. 30,000 status queries per month. ($0.25 per 1,000 transactions)|
Computer Vision API
It shows information about the visual content in an image. Developers can power their projects with this API to extract information from images to categorize and process visual data.
It can read text in images using optical character recognition (OCR), and extract handwritten text from notes, letters, essays, whiteboards, forms and other sources. Computer Vision API is also capable of recognizing over 200,000 celebrities from business, politics, sports, and entertainment. Additionally, it can identify more than 9,000 natural and manmade landmarks globally.
For example, for the above image, Computer Vision API will return tags like train, platform, station, building, indoor, subway, track, people, luggage, suitcase, etc.
|Pricing (Computer Vision API) :|
|Free||20 transactions per minute|
|5,000 transactions free per month|
|S1||10 transactions per second|
|0-1M transactions||$1 per 1,000 transactions|
|1M-5M transactions||$0.80 per 1,000 transactions|
|5M-10M transactions||$0.65 per 1,000 transactions|
Language Understanding Intelligent Services (LUIS)
It is used to develop conversational applications. Based on cloud API and machine learning, LUIS can infuse natural language into apps, bots, and IoT devices. It can identify required information from sentences in conversations.
Since it comes integrated with Azure Bot Service, developers can easily build sophisticated bots.
LUIS can learn actively to update and enhance the quality of natural language models.
|Pricing (LUIS) :|
|Free||5 transactions per second|
|Free Text Requests and Speech Requests|
|Standard||50 transactions per second|
|Text Requests||$1.50 per 1,000 transactions|
|Speech Requests||$5.50 per 1,000 transactions|
Speaker Recognition API
The apps built using Speaker Recognition API can be used to verify individual speakers or can be used as a voice authentication tool.
|Pricing (Speaker Recognition API) :|
|Free||20 transactions per minute|
|1,000 transactions free per month (Speaker Verification and Speaker Identification)|
|Standard||5 transactions per second|
|Speaker Verification: 0-50K Transactions||$5 per 1,000 transactions|
|50K-100K Transactions||$4.50 per 1,000 transactions|
|100K-250K Transactions||$4 per 1,000 transactions|
|250K-500K Transactions||$3.50 per 1,000 transaction|
|Speaker Identification: 0-50K Transactions||$10 per 1,000 transactions|
|50K-100K Transactions||$9 per 1,000 transactions|
|100K-250K Transactions||$8 per 1,000 transactions|
|250K+ Transactions||$7 per 1,000 transactions|
2. Amazon Lex
Amazon Lex allows developers to build conversational interfaces for applications using voice and text. The service comes powered by the same deep learning technologies that Amazon uses for its digital assistant Alexa.
Lex provides advanced deep learning capabilities like automatic speech recognition (ASR) and natural language understanding (NLU). The ASR enables conversion of speech to text, whereas NLU recognizes intent of the text. These capabilities enable development of applications with highly engaging user experience and real-life conversational interactions.
In computer science, speech recognition and natural language processing are few of the most complicated tasks. These technologies need deep learning algorithms to be trained on large amounts of data and infrastructure.
Amazon Lex eliminates these complexities by bringing the power of Amazon Alexa to developers.
Since it is a managed service, developers don’t have to manage infrastructure on their own. When the engagement of users increases on application, the developers wouldn’t have to worry about provisioning hardware.
The simple console that comes with Amazon Lex helps developers in the process of building a chatbot and infusing conversational interfaces into applications.
With the service, the voice or text chatbots can be published on mobile devices, web apps, as well as chat services like Slack, Facebook Messenger, and Twilio SMS.
Amazon has integrated the Lex with AWS Lambda, AWS MobileHub and Amazon CloudWatch. Further, developers can choose to integrate other AWS services like Amazon Cognito and Amazon DynamoDB. These integrations can boost the security of application and enable monitoring & user authentication.
|Pricing (Amazon Lex) :|
|$0.004 per voice request|
|$.00075 per text request|
Amazon Lex is available as a pay-per-use service. The charges are based on the number of text or voice requests processed by the bot.
TensorFlow is an open source machine learning framework that can be easily deployed and used across a wide range of applications. Google released the TensorFlow in 2015, and currently it is one of the most extensively used machine learning frameworks.
It comes with support for machine learning and deep learning, and allows easy deployment of computation, from desktops to clusters of servers to mobile and edge devices.
Google TensorFlow is an ideal solution for developers who want an AI platform that can lift heavy workloads and make AI projects from scratch. Developers can train their own image recognition system, and natural language processing models. The conversational AI chatbots can be developed with TensorFlow by training the models for specific data.
The TensorFlow ecosystem comprises a number of research projects and implementations to explore the role of machine learning in distinct use cases. For example, the Magenta by TensorFlow is a project that includes utilities to manipulate music and images to train machine learning models. This can be used to create new content from the models.
Developers can use TensorFlow to power their apps with numerous impressive AI capabilities. These capabilities can include the use of mobile camera to identify emojis, playing Pac-Man using images trained in browser, enjoying a real-time piano performance by a neural network, real-time human-pose estimation in browser, and teaching a machine to recognize images and play sounds.
Several leading companies including, Dropbox, eBay, Twitter, Uber, Nvidia, SAP, Intel, Qualcomm, LinkedIn, and more are already using TensorFlow to support research and production objectives.
|Pricing (AI and Machine Learning products) :|
|Google has defined the cost of training a model as $0.49 per training unit.|
|Predefined scale tiers- price per hour (and training units)|
|Basic- $0.2774 (0.5661)|
|Standard- $2.9025 (5.9234)|
|Premium- $24.1683 (49.323)|
|Basic GPU- $1.2118 (2.4731)|
|Basic TPU- $6.8474 (13.9743)|
4. IBM Watson
IBM is a well-known player in the field of AI because of Watson. It comes powered by modern innovation in machine learning to allow the models to learn more with less data. Developers can choose to build new models from scratch or use Watson APIs and pre-trained solutions to power existing applications.
What makes IBM Watson unique is its ability to learn from small data sets. IBM believes that it’s the quality of data, rather than quantity, which makes the difference.
IBM Watson comprises several enterprise-grade AI services, applications, and tooling. These include Watson Assistant, Watson Studio, AI OpenScale, Watson Discovery, Natural Language Understanding, Discovery News, Knowledge Studio, Language Translator, Natural Language Classifier, Personality Insights, Tone Analyzer, Visual Recognition, Speech to Text, and Text to Speech.
Additionally, Watson allows enterprises to integrate its services into Salesforce and Box. The Salesforce integration helps enterprises to provide AI-powered solutions and make quick and smart decisions across service and sales. Whereas, the Box integration is aimed to automate the structure of content, unlock hidden value, and automate workflows in the cloud.
|Pricing (Watson Machine Learning) :|
|Lite- Free||5,000 predictions|
|Standard||$0.50 USD /1,000 predictions|
|2,000,000 predictions included and then billed per 1,000 predictions|
|$0.40 USD /1,000 predictions|
Watson Assistant can be used to build virtual assistants for mobile devices, messaging platforms, and robots to transform the customer service department and more.
|Pricing (Watson Assistant) :|
|5 Workspaces, Each with 100 Intents and 25 Entities|
|Standard- $0.0025 USD /API call|
|20 Workspaces, Each with 2000 Intents and 1000 Entities|
|Premium- (price on request)|
|50 Workspaces per Instance, Each with 2000 Intents and 1000 Entities|
Natural Language Understanding is used to analyze text and extract metadata from content like keywords, emotion, sentiment, relations, and semantic roles.
|Pricing (NLU) :|
|1 custom model|
|Tier 1: $0.003/ NLU item for first 1-250,000 items|
|Tier 2: $0.001/ NLU item for next 250,001 to 5,000,000 items|
|Tier 3: $0.0002/ NLU item for next 5,000,001+ items|
Premium (Need to contact IBM)
Suggested reading: Comparing fog computing with edge computing
With several AI technologies and platforms available out there for building AI projects, it can be difficult to find the right one. Hence, it is important for developers and enterprises to thoroughly research multiple options before making a choice.
We have curated the top 4 AI engines which every enterprise or individual developer should consider while building projects or applications involving artificial intelligence in 2019 and beyond.
If you are already using an AI engine for your project, please let us know the review in comments below.