The Evolving AI Market
Artificial intelligence (AI) has worked its way into a variety of industries, from the obvious (autonomous vehicles) to the hidden (anti-money laundering due diligence). But while organizations are clearly recognizing the value associated with incorporating AI into their business processes, they also are encountering a number of challenges with integrating this new intelligence into their operational processes.
The value of using algorithms to unmask hidden patterns and then correlate findings with other seemingly unrelated variables to create real “intelligence” is becoming increasingly clear with each completed proof-of-concept (POC) project. But it is the larger, organization-wide deployment of AI that will generate the return on investment (ROI) that companies large and small have been seeking.
To fully access the operational and economic benefits of AI, however, organizations are realizing that, in most cases, enabling AI is not a plug-and-play proposition. Significant time, resources, and capital usually must be deployed, and in most cases, internal company teams are not experienced enough with AI, nor do they have the cutting-edge data science skills to adequately embark upon a truly transformational AI implementation journey.
Tractica analyzed nearly 300 different use cases to create forecasts across 29 different industries. Based on interviews with a variety of service firms, platform vendors, and tool companies, values for AI services were calculated for AI software installation, training, customization, integration, and support and maintenance, which cover the majority of tasks involved with an AI deployment.
Market Drivers
The market for AI services is likely to be largely driven by full-scale, enterprise-wide rollouts of successful POCs, which require far greater levels of integration, customization, and maintenance and support. The AI services market drivers are likely to differ somewhat from the drivers of AI software or hardware.
The key drivers of the AI services market include:
- A large and growing demand for enterprises to adopt and deploy AI technology across the organization
- Increasing complexity in the types of use cases in which organizations are seeking to deploy AI
- A lack of internal resources and/or technical expertise with AI technology
- A lack of experience or track record implementing successful AI solution rollouts
- The growing convergence of software and services
- An opening of the structure of the market for a more vendor and technology-agnostic approach to implementation
Each of these factors is working in concert with the maturation of the AI market as a whole. As AI moves from an experimental phase to one where real-world results are required, the demand for qualified assistance will increase.
Market Barriers
Despite the strong desire to incorporate AI into specific use cases, significant challenges still remain that can hinder the integration of AI, whether in a basic POC rollout or in a wider enterprise deployment. These challenges include:
- A reluctance to work with outside vendors and consultants
- Cost and scope concerns with AI projects
- Difficulty in getting project buy-in from senior executives
- Cultural and change-management concerns
None of these barriers are insurmountable, and, among companies and organizations that are serious about implementing AI technology, these concerns are likely to be viewed as being easily managed. Nonetheless, the concerns should be addressed before undertaking a large-scale AI project.
[This article is from research firm Tractica’s report on artificial intelligence services. View full report details.]
Key Services Offered
Most AI market services companies understand that it is no longer sufficient to simply offer AI software or AI tool sets to their clients. That is why smart vendors, whether software firms, consultants, or professional services firms, are either offering or partnering with firms that can offer a full suite of services to help support a large-scale AI solution.
The services solutions commonly offered by providers in this market include:
- AI Installation Services: These services include the actual development of a software solution, which includes the pre-installation consulting and planning work, as well as any POC or pilot project work.
- AI Training Services: Training services include any work required to identify, source, clean, label, or modify data for use by an algorithm. Work done to enable the actual training of an algorithm is also included in this category.
- AI Application Integration Services: Application integration services involve any work done to link or integrate multiple AI technologies together, as well as work completed to tie-in AI software with existing software or systems.
- AI Customization Services: Customization services cover any work done to modify AI solutions to meet the specific demands of a particular client or customer, and can involve the modification of the data inputs, the algorithm itself, the output stream of the algorithm or solution, and any customization of the tool or customer-facing interface.
- AI Support and Maintenance Services: Support and maintenance services involve any work done to ensure the AI solution is monitored, adjusted, refined, maintained, or repaired after the initial integration work has been completed. In many cases, a specific contract period will be used to establish roles, responsibilities, and timelines for delivering support and maintenance services.
While many of these services offerings are delivered as part of an overall solution, Tractica has segmented the tasks in the forecast to better delineate the actual process of delivering an AI solution.
Business Models in Use
While the exact structure and terms of an AI engagement will vary based on the provider, the client, and the use case, most AI engagements will follow a standard structure. Generally, projects are structured in a staged approach, starting with the research and selection of a use case. This stage of the process is critical, as a clear business case, along with stated goals metrics, must be decided upon to ensure the project remains on track.
Then, a POC program is initiated, which requires selecting data sources to feed into an algorithm, building or customizing a pre-built algorithm to the customer’s requirements, and then running and testing the solution to ensure it yields the expected benefits. Once the POC has been completed, which can take anywhere from about 3 to 4 months, the client and any services providers will evaluate the program, and decide whether to expand the program or whether to try a new use case.
If the program is expanded, the customer likely will engage the services of one or more outside vendors or consultants. While the fee structure and business models vary, generally speaking, providers may offer services on a project basis, a software license plus a consulting fee, a license fee plus additional fees tied to success metrics, or a combination of any of these models. The days of using a single vendor or consultant to handle an AI project are on the way out; services firms and software vendors are being asked to provide best-of-breed software, tools, and services, even if it means using open-source software or tools, or another vendor’s products and services.
Market Forecasts
Tractica projects that the AI services market will far outpace the amount of revenue generated by sales of AI software alone. Tractica projects that, on average, AI services costs will reach anywhere from 1X to 2X or more of the cost of AI software, due to the depth and breadth of solutions that are offered, as well as the extended time frame of delivery.
Tractica interviewed a wide range of platform providers, services firms, and technology companies for this report, and found that the additional costs of services are likely to vary widely. Nevertheless, several factors will impact the amount of a services engagement. Though not hard-and-fast rules, the consensus among market participants is that 9 times out of 10, services engagement costs are based on the following attributes of the use case and solution:
- Complexity: Intuitively, the more complex a solution is, the greater amount of work it will take to identify the data sources, clean and label the data, and develop the algorithm. Integrating AI into multiple back-end systems can add to complexity and require more work before, during, and after a solution has been put in place.
- Ubiquity: While AI technology is still relatively new, there are literally hundreds of successful solutions being deployed across a variety of use cases. Many of these are relatively simple implementations, but the learnings from these first- and second-generation solutions have formed a knowledge base for how to deploy AI technologies. Platform vendors, tool providers, and consultants have been able to “productize” some offerings, which then allows a somewhat more straightforward customization and integration process, which can help reduce services expenditures.
- Scale: Obviously, deploying a solution in the pilot or POC phase will require fewer resources and time than a full-scale rollout. But similarly, rolling out a solution that touches multiple back-end or front-end systems, or is expanded to multiple geographies or divisions, will encounter more complexity and cost.
Thanks to the advent of cloud computing, professional services will be concentrated in areas of the world where a significant amount of resources and will to develop AI technology abound, with services delivery concentrated in North America, Western Europe, and Asia Pacific. North America will dominate spending on AI services, with revenue reaching $88.6 billion by 2025, followed by Asia Pacific ($44.5 billion) and Europe ($43.6 billion). Growth of AI services in Europe and Asia Pacific will track closely, in terms of revenue, during the course of the forecast period, reflecting the global nature of AI technology.
