AI-as-a-Service: Getting SMBs Started with AI

​Human resource functions in small and midsize businesses (SMBs) are often on the outside looking in when it comes to using artificial intelligence to improve operations. That's because they typically don't have the resources required to build the internal infrastructure or hire the AI experts needed to make the technology work.

But there's a new option that allows SMBs to begin implementing tools such as chatbots with smaller upfront financial investment: Artificial Intelligence-as-a-Service (AIaaS) platforms. These third-party vendor solutions provide out-of-the-box AI capabilities to organizations in ways similar to Software-as-a-Service (SaaS) products now common in the HR technology industry.

Growing Gap Between AI Haves and Have-Nots

Recent research shows the gap in AI adoption between large and small companies continues to grow. A 2022 global study from IBM found that larger organizations now are twice as likely as smaller organizations to have deployed AI, compared with only 69 percent in 2021. Among the top factors hindering successful AI adoption are limited internal AI expertise, a too-high price tag, and lack of tools and platforms to develop AI models.

AIaaS platforms enable smaller businesses to begin using AI by providing ready-made platforms for developing and testing tools including chatbots that can answer frequently asked questions from job candidates and employees, or create application programming interfaces (APIs) that improve integrations between disparate HR systems. Using these pre-trained AI tools helps SMBs avoid costs of investing in specialized hardware such as GPU-based processers and other infrastructure needed to create and use on-premise AI.

In a time when there's a shortage of AI programming experts—who typically can demand hefty salaries—AIaaS platforms also enable SMBs to avoid those recruiting and compensation issues because most require minimal or no coding skills for successful use.

HR technology analysts say AIaaS platforms can be a good fit for HR functions in smaller businesses still figuring out how they want to use AI in their operations.

"AIaaS can be a good option for smaller or medium-sized businesses that don't have a big data science team or advanced infrastructure," said Alexander Wurm, a senior analyst specializing in enterprise software for Boston-based Nucleus Research. "These pre-trained AI models have taken off lately, whether for creating chatbots used on websites, for voice recognition purposes or for data classification. The platforms allow SMBs to see value right away from experimenting with AI with a low upfront investment."

Research shows more SMBs will likely invest in these types of platforms in the future. A study from Analysys Mason, a technology research and consulting firm in London, found that SMBs plan to increase their tech spending by 6 percent this year over 2022 levels, with some of that increase targeted to as-a-service technologies. Spending on Platform as-a-Service (PaaS) solutions, for example, is expected to grow by 18 percent this year over prior-year numbers, the study found.

"SMBs are demanding more flexibility without committing to expensive IT resources," wrote authors of the Analysys Mason study. "Early adopters are increasing their use of advanced analytics tools such as business intelligence and AI-powered applications."

AIaaS Providers Grow

While big cloud infrastructure players including Amazon, Google and Microsoft offer as-a-service platforms that typically include some AI capabilities, a growing number of smaller vendors have emerged to serve the AI needs of SMBs. Examples include vendors H20.ai, Levity and Odus.

These platforms are often priced on a per-usage subscription basis.

For example, pricing listed on the website of AIaaS provider Levity shows a monthly fee of $49 for smaller organizations seeking to automate simple workflows using AI. The fee covers 500 "actions" per month, with actions defined as any time AI is used to classify an image, PDF or piece of text (one e-mail body, for example). Every step in a workflow—such as sending an e-mail or updating a record—counts as an action. The monthly fee includes 10 AI "trainings" per month that involve teaching AI tools new tasks.

"There also are some platforms that allow you to import your own data and fine-tune an AI model based on that," Wurm said. "Those options are more expensive because you're actually training an AI model rather than just using it out of the box."

Lee Sustar, principal analyst with research and advisory firm Forrester, said other options for HR teams in SMBs include seeking out AI capabilities in technology platforms already being used in their organizations or even exploring options from larger cloud providers, which increasingly have industry-specific cloud products that can include AI capabilities.

"Some bigger players do have AI offerings targeted to SMBs or have partners who deliver those services," Sustar said. "For some midsize companies, AI capabilities also can be accessible in broader systems their organizations already use, such as call center or lead generation technologies."

Limitations of AIaaS Solutions

AIaaS platforms do have limitations HR leaders should be aware of, HR technology analysts say. Many platforms are designed for general purposes but can't easily be customized for use with company-specific data, for example.

"HR leaders should understand these platforms often can't be optimized for some of their specific use cases," Wurm said. "The platforms are easier to deploy and are lower in cost than other options, but you do see plateaus in performance and accuracy because you're not building the platforms yourself. For example, if you ask ChatGPT to do something really specific or niche, it likely won't perform as well because it doesn't have that level of specified training."

Brandon Medford, a senior principal analyst in Gartner's IT practice, said HR buyers should understand compliance, data privacy and security issues that come into play when partnering with AIaaS providers.

"For example, you'll want to know if there's any sensitive or proprietary data that the third-party provider will access, store or share that has compliance or regulatory issues," Medford said. "There also are the same data privacy and security concerns you'd have when partnering with any vendor where system integrations are involved."

Experts say HR buyers should ensure there's good visibility into how the AI in these platforms works to avoid the "black box" syndrome that can lead to artificial intelligence being used in biased or unethical ways when automating HR processes or making employment decisions.

Dave Zielinski is principal of Skiwood Communications, a business writing and editing company in Minneapolis.