In this article, you will learn how to:
- Turn customer AI fears into managed services contracts and position yourself as the trusted guide through deployment.
- Sell the “human in the loop” as a billable service.
- Build an AI practice that runs efficiently using the same AI tools you’re selling.
- Package the complexity of AI infrastructure and security as premium offerings.
- Get ahead of competitors in your market with knowledge and positioning strategies.
The human element
One of the critical components of any AI-based solution is the human in the process. Most experts in the field agree and have varied reasons for saying so.
This is a concern for many MSPs as well as their customers. The typical new technology fears about governance, risk, and compliance are broader. People are afraid that AI will take their jobs and replace them. Others fear AI could take over the company, the region, the country or even the world.
Even some of our greatest minds — including the late Stephen Hawking, Elon Musk and Bill Gates — long ago warned of the potential dangers of the uncontrolled advancement of AI.
Those who have watched a Terminator movie likely live in fear of the day Skynet becomes aware.
Why your clients need an MSP in the loop
Many MSPs have expressed a different concern. They want to know how to provide superior AI solutions to their customers. They have questions like:
- What exactly will they sell?
- Do they have the skills to sell it?
- Can they do so profitably?
- How do they learn all the terms, acronyms and other characteristics of AI?
- How do they position it with your customers?
- What is the MSP’s role in the onslaught of AI?
It’s clear that MSPs need strong grounding to build their AI initiatives. There are hundreds of thousands of people writing about AI today. Many are spreading wrong information because they’re ill-intended, uninformed or worse.
It’s important for MSPs to separate the noise and hype from the information that is most critical. It’s important to stay informed.
MSPs looking for deeper guidance on building, positioning and profitably selling AI solutions can turn to AgenticMSP. The Substack feature delivers practical guidance, relevant insight, timely news and actionable resources to help increase AI sales and profitability. It also helps MSPs reduce costs by leveraging AI in their own operations. Subscribe to AgenticMSP here to stay informed and evolve into an AI-focused MSP.
What humans and AI do and do not have in common
Many experts agree that humans are capable of things AI cannot yet process. Contrary to some opinions, we have not yet achieved artificial general intelligence (AGI), in which AI models can do everything humans can do. Some talk about artificial superintelligence (ASI), but we’re not there yet either.
There are many things we have in common with computers. We both have:
- The ability to store memories and information. We use our brains; computers use disk drives and solid-state devices.
- Input and output devices to send and receive information. Computers have keyboards, mice, microphones, speakers, scanners, plotters, three-dimensional printing capability and more. Humans have our eyes, ears and skin. We even have one input device that computers don’t yet have: our noses.
- Processors. Nvidia, AMD, Intel and others keep innovating and improving the computer processors running AI. Humans have the processor in our brains, said to reside in the prefrontal cortex.
An instruction set architecture (ISA) is the complete vocabulary of commands that a processor understands and can execute. It is essentially the contract between hardware and software. The ISA defines four core things:
- Instructions: The actual operations the CPU can perform (ADD, LOAD, JUMP, COMPARE, etc.)
- Registers: Small, fast memory slots inside the processor where actively used data is held
- Data types: The kinds of values the processor recognizes (integers, floating-point numbers, binary strings)
- Memory access model: How the processor reads from and writes to RAM
With the abilities available in computer processor instruction sets, we have enabled them to add, subtract, multiply, divide, analyze patterns, make simple decisions and generally work in many ways with data.
All these abilities reside within our prefrontal cortex as well. We can do all those things, albeit slower.
Howard M. Cohen
The ‘mystery instructions’
Inside the human brain’s ISA, you can also find things like instinct, intuition, imagination and emotion. These instructions allow us to access and take advantage of certain resources.
Computer processors do not have this information, which is passed down to us in DNA. No one has yet duplicated these instructions in a computer processor’s ISA. In fact, nobody has even articulated what they are or how they work. That likely won’t change anytime soon. Until it does, humans will have abilities beyond those of any computers.
Why AI needs you as much as you need it
Human-only abilities are critical, and computers running AI software require access to them. The only way for them to benefit from these unique abilities is from interacting with humans.
That’s why we always need to keep a human in the loop in all AI-based initiatives.
AI brings several advantages to the relationship with humans. The technology can access unimaginable quantities of data through large language models (LLMs) almost instantaneously. Humans can benefit by asking questions and letting the LLMs provide the answers.
Prior to AI, computers could only distinguish between a one or a zero, an on or an off. Over time, information became decipherable and manageable by connecting long strings of ones and zeroes to represent the characters we humans use routinely.
We began to achieve artificial intelligence with the development of neural network technology. The neural network’s core advantage shifted computers from following rules to learning patterns. That single shift unlocked capabilities that decades of traditional programming could never achieve.
With pattern recognition, computers could go beyond the simple zeroes and ones and detect and work with nuanced “shades of meaning.” This ability soon resembled human behavior but still lacked those mystery instructions. A human is still required to interpret many of the observations and calculations that AI delivers. A human often must ensure that calculations with less-than-definitive results could be properly managed.
The importance of keeping the human element
You’ve been doing this forever. Your customers love the things their computers can do for them, but they don’t want to have to worry about their ongoing operations. When that stops, they call upon you, their MSP.
Expect this to extend into how you deliver AI solutions. “Easy-to-use” has always been easy to say. Users require guidance and sometimes assistance in developing the ability to productively use new technologies. They will continue to need their MSP to keep everything running, including the new software that has created a dramatic increase in what they can do with AI.
AI systems have the same infrastructure requirements as any system, and then some. We’re only now beginning to understand the complexity of securing and protecting AI systems that are moving data across more environments than ever before. That complexity hinges on many disparate systems working harmoniously together.
FAQs
Q: What exactly am I selling when I sell AI?
Managed AI services. It’s the ongoing guidance, oversight, infrastructure, security and human interpretation that AI deployments require. You’re not selling a product but the expertise and continuity that makes AI work reliably for your customers.
Q: Why do my customers need me if AI does everything automatically?
It doesn’t. AI lacks instinct, judgment and contextual understanding. Someone must interpret its outputs, catch its errors and make the calls it cannot. That’s a billable, recurring role.
Q: How do I explain AI’s limitations to customers without undermining the sale?
Lead with what AI does brilliantly: pattern recognition, data processing and speed. Then, position your role as the essential complement. AI is powerful, and power requires management.
Q: What new infrastructure does AI require?
More data movement, endpoints, interdependencies between systems and a much larger security surface. It’s everything you already manage amplified. That’s more contracts, not fewer.
Q: How do I price AI managed services?
Start with what you already charge for comparable complexity. Then, add on for the monitoring, governance and human-oversight components that are unique to AI. This is emerging territory, so early movers set the pricing norms.
Q: Do I need to become an AI expert before I can sell this?
No. You need to be ahead of your customers, not the researchers. Gain enough knowledge to guide, govern and troubleshoot. You build expertise by doing more.
Q: How is this different from what I already do?
The underlying service model is familiar. You keep things running and secure, and serve as the expert your customers rely on. The technology is new; the relationship isn’t.
Q: Where do I start?
Begin with the customers who are already asking about AI. They’re your pilot engagements. Use them to build your practice, pricing and confidence before you go broad.
Since 1981, “Senior Resultant” Howard M. Cohen has managed sales, marketing and business development for channel companies. He then ran technical, professional and managed services businesses, including one of the nation’s earliest MSPs. He now publishes the AgenticMSP Substack. You can reach Cohen at hmc@howardmcohen.com.
Featured image: peno — stock.adobe.com
