Classifying The AI Vendor Landscape: Making Sense Of The Market

Classifying The AI Vendor Landscape: Making Sense Of The Market
Classifying The AI Vendor Landscape: Making Sense Of The Market

According to Crunchbase, there are now over 8000 vendors in the AI marketplace. These vendors cover the gamut from data science platforms enabled with machine learning capabilities to marketing automation tools that help to optimize advertising and everything in between. The problem is that without some sort of organizational framework, it’s hard to understand how these various vendors relate to each other or can be used in conjunction with others. If you’re an enterprise end-user or AI implementer trying to pick and choose vendors to work with, this is more than just a small problem!

The existing diagrams of the vendor landscape are a confusing mess of logos piled together into a graphical logo salad that doesn’t really make much sense. In one corner, you have machine learning platform tools, and right next to it are tools geared for financial assistants. Other than the placement of being next to each other, these tools and applications share little else. There’s no explanation of what those logos are, how they’re related to others in the same bucket, and how those buckets relate to others. Often these vendor diagrams are nothing more than a simplistic way of saying “hey look at how much is going on in the market”. But with over 8000 vendors marketing AI wares, this way of trying to understand the market leaves much to be desired.

Late in 2018, AI research and advisory firm Cognilytica published a comprehensive analysis of the market. As part of this, they released a classification of the AI vendor landscape with 4 major layers, 40 primary categories, and 100 subcategories. Through this classification, organizations can more clearly understand how the AI market is organized, make clearer comparisons between vendors, and gain insight into how different layers of the AI vendor stack relate to each other.

Like many technology ecosystems, the AI vendor landscape can be arranged into a series of layers, where one layer of the technology enables and powers the layers above it. Like a cake, every layer of the vendor landscape stack is tasty. No one layer is “better” than another, just like every layer of german chocolate cake is as yummy as the next. But like a cake, each layer depends and builds upon the layer below it. In the same way, Cognilytica’s AI vendor “layer cake” shows each layer provides significant value to the AI picture, and one layer depends and builds upon the layer below it.

An overview of the hierarchy is here:

Cognilytica has four primary layers of the AI vendor ecosystem:

AI & Machine Learning Infrastructure – Technology offerings that are not necessarily AI or ML specific, but are incredibly helpful or even necessary to enable AI projects and solutions.
AI Enabling Technology – Solutions specifically built to enable AI, machine learning, and cognitive technology projects, but aren’t built for a particular application of that technology
Horizontal Applications of AI – Cognitive technology-powered solutions that provide a particular application, but not in a manner that is specific to an industry or domain
Industry- and Domain-Specific Applications of AI – Applications that are particularly enabled by AI technologies and applied to a specific industry or domain problem.

These layers are further defined below with examples of what’s in each layer.

THE AI & ML INFRASTRUCTURE LAYER: At this layer, technology solution vendors provide offerings that support the broad range of underlying infrastructure needed for cognitive technology projects. These solutions aren’t necessarily purpose-built for AI and ML, but they provide necessary underlying infrastructure technology from vendors that specialize in that infrastructure. While you can use this infrastructure to help with your AI projects, you don’t necessarily need to, as they are more broadly applicable to a wide range of applications and use cases.

In this layer, we find computing infrastructure such as GPUs, cloud computing capabilities, and specialized chipsets. At this layer we also see the range of big data solutions and providers that have added AI and ML capabilities to their platforms. Likewise, data science solutions that have added ML specific capabilities exist in this layer. Cognilytica’s research uncovers around 300 vendors at this layer of the stack.

THE AI ENABLING TECHNOLOGY LAYER: At the next layer in the AI vendor stack, we have AI enabling technologies. These purpose-built AI technologies are meant to help provide core cognitive technology capabilities. These solutions are specifically adapted for AI, ML, and cognitive technology applications and aren’t really applicable or suited for non-AI applications. As such, these are truly technologies that enable specific AI and ML capabilities, which others can then build their own AI and ML applications on top of.

At this layer we find natural language understanding solutions, computer vision enablement, content intelligence, machine learning platforms for development and operation of ML projects, autonomous vehicle infrastructure, and advanced robotics enablement. Cognilytica’s research identifies around 700 vendors at this layer of the stack.

THE HORIZONTAL APPLICATIONS OF AI LAYER: At the next layer in the stack, we find solutions that are purpose-built to solve particular problems with the use of AI. These offerings are not aimed at a specific industry or class of industry problem but solve a class of problems that share some common application. For example, voice assistants are a horizontal application of AI because they are a particular implementation of AI for a particular problem, but not one aimed at a specific industry. However, while these horizontal applications have broad applications, they can’t be easily repurposed for some other horizontal applications. For example, we can’t use voice assistants to navigate autonomous vehicles or cognitive automation tools to intelligently process documents.

In this layer, we find vendor solutions that provide operations intelligence, predictive analytics and decision support, conversational systems (including voice assistants), intelligent document processing systems, task assistants of all sorts, and complete autonomous machines and vehicles (not just enabling technology). Cognilytica’s research identifies around 1500 vendors at this layer of the stack.

INDUSTRY AND DOMAIN SPECIFIC APPLICATIONS OF AI: At the top layer of the stack, we find AI and cognitive technology solutions that are tailored to the specific needs of industries and application domains. These solutions are not meant to enable you to build other solutions on top of their offerings, nor do they provide general AI and ML infrastructure or horizontal solutions applicable to a wide range of industry or domain applications. Rather, these solutions leverage AI and cognitive technologies as the key element to enable their solution functionality. It should therefore be no surprise that this top layer of the AI Vendor Classification is extremely diverse, with dozens of further categories and distinctions. In fact, it would be impossible to truly identify every domain-specific application of AI. The entire reason why AI is transformative is precisely that it is impacting so many industries and domains.

The categories identified in this layer are meant to be illustrative of the wide range of specific solutions that AI can enable. Some of these solutions are for industries that are generally accepted as “verticals”: finance, healthcare, insurance, energy, and utilities. But specific “domains” are also identified where the application is to a particular problem area. This includes cybersecurity and physical security applications, sales and marketing, news and content production, education and knowledge management, and the like. Cognilytica’s research uncovers a remarkable 3000+ vendors (and growing) at this layer of the stack.

MAKING SENSE OF THE MARKET: As you go up the stack, the diversity of the solutions expand as do the total number of vendors. This should not be surprising. There only needs to be a small set of AI infrastructure providers that can, in turn, be built upon by a larger set of the enablement technology platform and vendor providers, that can, in turn, enable an even larger set of application use case vendors, that in turn enable a gigantic number of specific applications. This is why, for example, Amazon’s Alexa device platform exists at the Horizontal Applications layer of the stack, and the custom skills built on top of the Alexa device sit at the Industry-and Domain-specific applications layer. The Alexa device in turn sits on top of the natural language enabling technologies, which then rely on computing, storage, and big data infrastructure to operate.

By grouping vendors in a more logical manner than simply throwing a bunch of logos together into a chart that puts all the different applications, use cases, and technology infrastructure layers at the same level of the hierarchy, you can gain a better understanding of how the market is sorting itself out. As the number of vendors in the market continues to climb, classification and organizational charts of the type provided here will continue to show their value.

originally posted on by Ron Schmelzer