Just jump in.
That’s the best advice Ben Schreiner, head of AI, modern data and modern data strategy business development at AWS, has for SMBs looking to get into AI.
“We encourage everybody to get started. Everybody is in a learning phase. And so, whether you’re using it personally or starting to use it professionally, getting started is certainly the most important first step,” Schreiner said on the latest episode of the Targeting AI podcast from AI Business.
But AI competency matters. AWS — and other big tech vendors — offer multiple programs to help SMBs get up to speed with AI.
“Some of the findings from our research are that, indeed, SMBs are in a hurry,” Schreiner said. “They see this as a competitive tool that they want. Partnering with someone who’s done this before and can understand your business and what you’re trying to do is a great way to accelerate.”
Schreiner also encouraged SMBs to start small and avoid overly ambitious projects. While he acknowledged that SMBs are facing all kinds of challenges, including skills gaps, cost unpredictability and mounting operational complexity, he said he sees the biggest blocker as a lack of data quality.
“An important thing to do is to realize that AI will only be as good as the data it has access to,” he continued. “SMB folks [say] they don’t necessarily trust their data. AI isn’t going to magically make their data better.”
Another barrier is AI overkill. Choosing the right tools for the job is key.
“If you have the biggest model and you’re throwing all your problems at it, you’re solving teeny tiny little problems with this great big model,” Schreiner said. “And that is a return on investment mismatch. Have the right model for that problem.”
AWS offers some open-weight and open source AI models that are good for smaller operations. But the vendor — like other tech giants and their AI platforms — also gives customers the opportunity to choose from a variety of third-party models, including powerful frontier models.
“They’re all good at different things,” Schreiner said. “What you’re trying to do is use the model that’s best suited and best priced for what you’re trying to do. So, you get that match, and you’re not spending a lot of money to solve very small problems.”
