What MSPs Should Keep in Mind as They Deploy Agentic AI

MSPs are either already fielding requests from clients to include agent-based AI into their workloads, or they will soon be seeing those requests. We delve deeper into the possibilities and potential pitfalls of deploying agents with AI Enablement Principal Derek Ashmore, who works at an advisory firm specializing in AI adoption.

GenAI and AI agents still aren’t perfect– and neither are humans

Ashmore’s role focuses on how tech partners and users more generally adopt and leverage various AI tools. To him, as he continues to learn more about how companies unlock value through AI adoption, the most important thing to keep in mind is realistic expectations.

No AI tool, whether it is a GenAI-based chatbot or the emerging agentic AI solutions now available, can handle everything a company does without oversight.

“AI is not perfect. I like to say I have a team of agents that does work alongside my human team, and sometimes the agents go astray and need coaching and correction,” Ashmore said. 

Ashmore also notes that it’s essential for MSPs to recognize that for many of their clients’ workers, AI tools represent a new skill set. The natural language-based prompting and ongoing development of the ability to use the tools will be unfamiliar territory to some, which is why training remains critical for businesses seeking long-term ROI from their deployments.

Mitigating security risks inherent in AI and all technology 

Ensuring AI is deployed securely is another way for MSPs to support their clients’ technical and business needs. 

There are many security concerns within AI solutions, including:

• Controlling sensitive data and ensuring AI solutions only access data they should
• Regulatory compliance as governments and other bodies address AI technologies
• The likelihood that tools such as chatbots can be jailbroken and accessed
• AI-powered ransomware and other attacks, as threat actors also leverage generative and agentic AI in their workflows.

To Ashmore, while the above is essential to keep in mind, it isn’t really all that different from what partners and their clients should be doing anyway.

“I challenge anyone to name a technology that someone couldn’t, and hasn’t, taken and misused for harm,” Ashmore said. “Just like any tool, there are steps you need to take and guardrails you have to put into place to mitigate risk for you and your clients.”

AI changes roles, but it doesn’t eliminate jobs: why adaptation is key

Ashmore also thinks AI is a game-changer, but not a job stealer. The general public remains wary of how the technology will impact the market, and partners themselves have occasionally voiced concerns that AI demand in the market has thrown the market into a flurry of change.

“I see all sorts of nervousness about AI taking jobs away. What I will say is that AI will definitely change jobs, but it won’t take a job away from you unless you refuse to adapt,” Ashmore said.

Part of his confidence in that stems from the experience he has had in focusing agents on the work his staff doesn’t want to spend their time doing. As we have reported several times over the past few months, many partners are finding success in automating mundane and routine tasks that technicians have often disliked.

“I make them do the things I have trouble getting humans to do,” said Ashmore. 

Constant testing and shorter decision horizons help MSPs stay ahead of AI’s rapid evolution

For some, the sheer amount of innovation within the AI space feels too overwhelming to stay on top of, let alone get ahead of. To Ashmore, the feelings are understandable, but he says everyone can, and should, take the initiative to understand the technology anyway.

“You still have to understand the technologies you’re working with,” Ashmore said. “I am entirely self-taught, and I have always said that I learn for a living. I’m struggling to keep up, too. I feel very much like the world is ahead of me and I’m running to keep up, but to some extent, we have to get used to that.”

He says his strategies for keeping up with the moving target include constant testing and trials with new tools as they enter the market. He also encourages his team to do the same and share their learnings, as well as using AI tools themselves to surface new ideas and opportunities.

Ashmore also stresses that MSPs and other providers must act quickly and continue to adapt and evolve, or risk being left too far behind to recover fully.

“Time isn’t what it used to be, honestly. Things are just moving much faster than they ever did,” Ashmore said. “The decisions that partners make now need to be on a shorter-term basis. In this world, you can’t look at a five-year horizon anymore.”