Do You Know Which Platforms Power AI Services?

Like any software, Artificial Intelligence can be packaged and wrapped in a number of ways from being embedded in an app, programmed into a robot, or sit in a chip including other applications. But which platforms do they play in that allow them their ability to learn, automate and make decision? Let’s take a look at existing and custom build AI platforms.

Existing AI Platforms vs Custom AI Platforms

Existing Platforms

Below are some examples of existing AI platforms that power everyday user services through robot, mobile, voice or other mechanism, and across every sector.


TensorFlow™ is an open source software library for high performance numerical computation. [1] It was released as an open source framework during Google’s Open I/O 2015, the Google created software allows developers to harness the system for their required use. It is reported that Tensor Flow has become the world’s most widely adopted machine learning framework. [2]

It was announced at the TensorFlow Dev Summit 2018 that the software hit a milestone of 11 million downloads. TensorFlow representatives also talked about how they are bringing TensoFlow Lite to Google Apps by working across other Google teams. And DeepLearn JS
introduced their new partnership with the TensorFlow family. Rajath, Director of Engineering at Tensorflow talked excitedly about how the popularity of TensorFlow has grown over the last two years. “There are over a 1000 contributors to the TensorFlow code from outside Google. And the new TensorFlow Blog new newly launched will share work by the team and the community and invites participation from the global community too.”

Google Cloud AI

Google Cloud AI promotes the products as delivering secure, open, intelligent, and transformative tools to help enterprises modernize for today’s digital world. [3] Among the many products that Google Cloud AI offers for various applications, TensorFlow is the backbone of the Cloud TPU’s offered by Google. Some of the other service offered include: CloudAutoML, Cloud Machine Learning Engine, BigQuery ML, a Cloud Talent Solution, among many other services.

In latest developments, Diane Green who became CEO of Google Cloud Services two years ago when her starutp was acquired, will be replaced by former Oracle executive, Thomas Kurian. It is reported that Google Cloud Services is not catching up enough to overtake Amazon Web Services who have the majority market share in ‘shift and lift businesses.

And a few hours ago, Google Cloud announced expansion of its partnership with global cybersecurity company Palo Alto Networks to simplify security and accelerate cloud adoption. The company will run its Application Framework on Google Cloud to take advantage of Google Cloud Platform’s AI and analytics tool. [4]

Amazon Web Services

Amazon Web Services or AWS tells potential customers that AWS has the broadest and deepest set of machine learning and AI services for your business. [5]

They currently have the highest market share in cloud services at 51% leaving Microsoft, Alibaba and Google in 2nd, 3rd and 4th place respectively.

AWS giant market share against competitors

AWS has also followed Google in to the AI chip market by releasing their own chip. The Inferentia chip will become available to customers late next year. The Chip was announced at the company’s AWS Re:Invent user conference in Las Vegas; “The chip will provide A.I. researchers ‘high performance at low cost.’” AWS has said that Inferentia will only deal with inference in machine learning for now, and not running predictions. Their reasoning for introducing an inference only chip for now; Cost-Effectiveness. The company also said that customers will be able to use Inferentia with TensorFlow AI software (created by Google), as well as other AI frameworks like PyTorch and the ONNX format for converting models.

Google and Alibaba were the first companies to announce AI chips for consumers and developers to include into products.

Microsoft Azure

Microsoft Azure promises that “with the flexible Azure platform and a wide portfolio of AI productivity tools, you can build the next generation of smart applications where your data lives, in the intelligent cloud, on-premises, and on the intelligent edge.” [6] Some of the services include: image classification with convolutional neural networks; information discovering with deep learning and natural language processing; defect prevention with predictive maintenance; enterprise productivity chatbot, and many more solutions.

In the latest news Microsoft’s Machine Learning Service is now generally available to developers and data scientists who are looking for efficient ways to build machine learning models. This comes days after Amazon Web Services announced enhancements to its Sagemaker. [7] The company is looking to keep up the pace with Amazon and Google.


Offering Data, Analytics, and AI as one of their key service offerings, the company’s HOLMES is Wipro’s artificial intelligence and automation solution platform. Wipro says that its AI-First Framework is designed to enable data monetization and encompasses: Robust processes; Next-gen technologies, and leading-edge capabilities.

During November 2018, the HOLMES Advisory Board was set up.

Rohit Adlakha, Wipro’s Global Head for Wipro HOLMES™ and Automation Ecosystem and the man behind HAB says “The objective is to enable a forum where leaders can exchange ideas, experiences, and best practices in terms of practical insights on applied AI usage in enterprise contexts in a non-commercial setting. The forum brings together three key drivers – AI’s potential is being advocated by thought leaders, tools are built by product/ solution providers, and use cases are being implemented & used by enterprises. All three learn from each other on an “as-needed” basis, but more often than not, the interactions between them are one-on-one”. [8]

Infosys NIA

Infosys launched NIA last year: The Next Generation Integrated Artificial Intelligence platform. The system was released having been built on the company’s first generation AI platform Infosys Mana. Infosys says that the company manages big data/analytics, machine learning, knowledge management, and cognitive automation capabilities of Mana; end-to-end RPA capabilities of AssistEdge; advanced, high-performance, and scalable machine learning capabilities of Skytree; and optical character recognition (OCR), natural language processing (NLP) capabilities and infrastructure management services. [9]

Infosys NIA Framework

Infosys released the results of an analytics research poll at the beginning of December 2018 which they conducted with 1,062 senior executives at companies with revenues exceeding $1 billion. Respondents representing 12 industries were based in Australia, Europe, and the United States. [10] The analytics poll notes that data analytics tools are being augmented by AI, the Internet of Things and cloud technologies as users look to move beyond getting their arms around data to using voluminous data sets to boost revenues.

Custom Built Platform

These examples simply represent custom built platforms, that can be built on top of the above services that are listed i.e. Google Cloud AI, TensorFlow, etc. We’ve included video explanations that give you a practical example of the system’s output, either using predictive analysis, voice recognition, and/or automation.

SAP Co-Pilot

AthenaHealth uses WalkMe

Which works best for your business?

Dependent on your organisation’s need and business intelligence data that is collected, your business can manifest AI application in several ways. Firstly identify the problems that you want AI to solve. [11]

  1. How would you add AI capabilities to your existing products and services? There should be specific use cases that you’d like to solve for customers. Automating the customer care process. Using predictive analytics. Image recognition as part of the security or buying process etc.
  2. Acknowledge the internal capability gap. This means identifying resources both human and software in the organisation that can carry out the AI capabilities you would like to implement.
  3. Bring in expert skill sets and create a pilot project as a test.
  4. Begin the process of internal data integration. Your company or organization wants clean data to enable better machine learning. Good storage ability is also key.

The above suggestions implies using custom built software to serve a particular need, like assessing insurance risk on an applicant.

If your business is looking to implement an AI layer by starting with affordable and realistic goals, then automation is the first step. From implementing a chatbot on social channels and the website to using a tool like Gong to analyze sales performance, and using the targeting options offered by Facebook’s AI advertising system, there are many instances of implementing AI as a smooth and simple first step.


[1] TensorFlow, About Tensorflow,

[2] Tech Includes, “TensorFlow 2.0 announced, here is what you can expect”,

[3] Google Cloud AI, Why Cloud AI,

[4] ETCIO, “Google Cloud expands partnership with Palo Alto Networks”

[5] Amazon Web Services,

[6] Microsoft Azure AI,

[7] Tech Target / Search Cloud Computing, “Microsoft ups its AI game with Azure
Machine Learning service “,

[8] YahooNews, “HOLMES Advisory Board set up to boost AI adoption and foster collaboration”,

[9] Infosys, Infosys Launches Infosys Nia™

[10] Datanami, “Analytics Use Grows in Parallel with Data Volumes”,

[11] PC Mag, “10 Steps to Adopting Artificial Intelligence in Your Business”,