Small- to medium-sized businesses (SMBs) want to adopt AI for efficiency – whether to reduce manual tasks, improve customer experiences or streamline processes. However, they don't have the resources – both people and budget – that larger businesses have. AI spending is also increasing rapidly. According to Gartner, AI spending is up 18 times from two years ago.
TechTarget sat down with Matt Meyers, executive director and general manager, North America SMB at Lenovo, to discuss how SMBs can implement AI into their business processes. Meyers describes how IT leaders should prepare for implementation, including data cleanliness, security and governance policies and strategies to work with outside vendors. He also shares the common problems SMBs face during implementation and how leaders can avoid them.
Editor's note: The following interview was edited for length and clarity.
What should SMBs do to prioritize and prepare for AI implementation?
Matt Meyers: For SMBs, a lot of my AI conversations and my team's conversations relating to AI tend toward: Where does your data live? Where are your ingestion points for that data? How are you orchestrating and managing that data? Data is the most critical and functional component of implementing AI to maximize your business priorities. Once you have the data in the right place, security governance is obviously critical, and there are many capabilities.
There's a lot of focus on security right now as it relates to AI, and the next piece is: What do you want to do with that data within your security and compliance restrictions, or given your capabilities? And typically, the conversation goes in two different directions, and they can come back together at different points. But it's about this: Do you want to build productivity and automation into your business, or generate more revenue from the data you already have?
How can main IT teams simplify technology management?
Meyers: From a technology management standpoint, whether you're talking about IT asset management or IT service management, it can become very complex very quickly. Many teams that I tend to speak with try to break it into silos, whether it's cost or security or risk or compliance, whatever it may be. The beauty of it is that we really shouldn't be siloing those pieces. They're all critical focus areas, but we can identify clear business outcomes that we're trying to achieve.
Lean back on the building blocks of implementing AI in very practical ways. For the organizations I chat with, I lean into the programs that have been established to make AI practical in the end. We try to avoid the siloing of different AI projects or components and really bring them together around the business outcomes those organizations are trying to achieve. Because again, a lot of SMBs don't have huge IT teams. It's usually one decision-maker and three or four support people, or whatever it may be; bringing those individuals into a room and addressing all those components, with the business objectives front and foremost in the conversation, for the practical implementation of AI.
What are the biggest IT complexity challenges that SMBs face today compared to larger businesses?
Meyers: Security and cost are absolutely the two biggest challenges for SMBs compared to large enterprises. Large enterprises may have similar issues, but to a large degree, they have more flexibility. They have larger budgets and can take money from one place to invest in another. They have larger IT teams. Cost and resources are two things that not all SMBs have. By maximizing cost, we've seen the implications in our macro ecosystem of the daily rise in storage and memory costs.
But more and more, we're seeing that trend in the IT ecosystem around Opex consumption [budgeting that emphasizes operating costs rather than large upfront purchases]. A lot of organizations are used to cloud consumption models, which are naturally Opex. So, as they're evaluating hybrid cloud solutions, bringing some of those workloads and use cases on-prem and building out that data center, whether it's on-prem or in a colocation facility.
“Security and cost are absolutely the two biggest challenges for SMBs compared to large enterprises. Large enterprises may have similar issues, but to a large degree, they have more flexibility.”
Matt Meyers
executive director and general manager, North America SMB at Lenovo
How can companies decide which legacy systems to retire, replace or increase?
Meyers: There's a place for everyone to use lots of different technology and capabilities. Our cloud partners are critical to how Lenovo goes to market, and so whether it's Google Cloud, AWS or Microsoft Azure or whatever cloud provider it may be, those are great platforms, but they're especially relevant in the future when we get into the conversation around tokenomics for the spiky workloads – the things that aren't being tapped into all the time. They're great for blue-sky development, but not all workloads will have a place there in the future because of the cost of tokens required to run AI platforms.
So, the next piece is for those steady-state workload use cases. What is the utilization? Usually, our recommendation is to put those steady-state workloads on-prem or in a colocation, somewhere where latency, redundancy and security are within their control. And then the next piece we talk about is NPUs on the latest generation of laptops, desktops and workstations. There are many cases where, compared to two years ago, organizations can develop small language models to run on these NPUs, independent of Wi-Fi connectivity or whatever else, for quick, individual access to AI. It's kind of that roadmap we've built out, around putting the right workloads and use cases in the right places for longevity and sustainable performance, not just the easiest access to put something wherever it may lie.
What foundational IT capabilities does a business need in place before adopting AI tools effectively?
Meyers: Data cleanliness is absolutely critical, and it is hard. Data cleanliness for any organization, whether it's an SMB, a large enterprise or the public sector, is about being able to sift out the right data to use. This is one of the biggest challenges for any organization, but especially for small and medium businesses. Getting the data right is foundational to using AI.
The second piece is ensuring you have the right data governance, compliance and security practices in place. If there's anything I would recommend, and I do on a daily basis, to small and medium businesses, it's that we have to start with the data, the cleanliness of that data, and the security aspects of that data before we ever start building out the AI capabilities. Because the risk of not cleaning your data is very real. You can train these AI models with bad data, and then you get negative outcomes from that. And so, data compliance, data governance and security are the two foundational pieces above everything else that I chat with customers about every day.
How should companies think about data privacy and compliance when getting internal data into AI systems?
Meyers: It is a major topic of discussion with almost every single customer. There's a lot of news articles out there about using public-source AI tools that have a lot of capabilities, but feeding your own intellectual property into public-source AI platforms represents massive risk. And so, we map out many of those capabilities and risks from both a platform development and an implementation standpoint. Because one of the hardest pieces, one of the biggest vulnerabilities to any organization, is the human element of using AI, and you have people who can make mistakes or may just not be thinking about what they're putting in these public spaces. So, the reality is also cultural enablement of AI within organizations.
With limited budgets, how should SMBs prioritize between simplifying existing IT and investing in new AI-capable?
Meyers: Maximizing AI productivity within any organization is oftentimes a topic of debate, right? Especially in the macro-IT ecosystem right now, there are many blue-sky initiatives we can pursue. There are a lot of steady-state pieces we can help improve productivity around, so the rationalization we always point back to is: What true business outcomes are we trying to achieve? And a lot of times we talk about, if you're running agentic AI in your ecosystem, they may be identifying problems that you didn't even know existed before, which is a fun conversation to have with customers because then you get into the conversation of real issues versus, if you want to call them, ghost issues. What are we training these AI models to do to drive performance within our business? A lot of times, if the data compliance piece isn't right up upfront, you could be training these AI models the wrong way. They could be providing the wrong insights regarding your business outcomes, so structuring things appropriately around data governance and security upfront is critical. But driving to those business outcomes is where we're really focused – simplifying and modernizing IT is the foundational layer.
What's the biggest mistake you see mid-sized businesses make when trying to modernize IT for AI readiness?
Meyers: Yeah, too many blue-sky initiatives. I love discussing blue-sky initiatives with SMBs, when it comes to AI, it comes down to practicality. Every aspect of a business will have individuals coming forward asking to use AI. But the best organizational implementation of AI typically has a steering committee or governance model, and a center of excellence for AI to help filter good ideas and move forward with the great ideas that best align with a business's objectives. And so again, it relates back to how we drive the most value, the most ROI, out of these ideas versus just focusing on trying to make a million blue-sky ideas around AI, implementing things that will align with the business objectives.
What's the one piece of practical advice you would give an IT leader starting this journey?
Meyers: Starting this journey, I would say engage with those who are willing to partner with you early. You don't have to do it alone. There are many capabilities in this industry. We work with small and medium businesses every day, focusing on those key components again, particularly data compliance and security. Focus on that layer significantly. Partner with those that want to engage with you. There's value in being able to work within your own teams and prioritize those business outcomes. Once you've gotten to that point, don't wait to engage a partner. Engage them early. Let them help you be part of that process.
Amanda Hetler is senior editor and writer within the Reader Engagement and Outreach team of Informa TechTarget.
