Develop Your “AI Assistant” Training First

Many MSPs tell me they don't know where to begin with AI. I'm a believer in the crawl, walk, run, fly approach, so I recommend they begin by getting their customers comfortable with AI basic uses

Apr 04, 2026

∙ Paid

Please Note: Once again this is a particularly long post, so subscribers will find the emailed version of this issue cut off at the end. Please go to the web version for the complete version.

If You Find AgenticMSP Valuable: Please recommend AgenticMSP to your friends and colleagues. Also, please offer up your comments, observations, questions, or any other responses in the chat below. Let’s get a hearty conversation going!

Develop Your “Assistant” Training First

Before you can sell anyone a solution, you must first prove they have a problem. Before you can sell anyone anything, you must first prove that they can use it and use it productively!

All the Talk About AI Assistants

As you read articles about AI, including those I include regularly here in AgenticMSP, you’ll keep running into talk about AI Assistants.

· An article in huckster.io from two years ago is titled “NVIDIA’s Jensen Huang Predicts a Future Where “Everybody Will Have an AI Assistant“

· DeepMind CEO Demis Hassibis envisions “a universal assistant... so valuable it becomes an everyday companion, one that understands you, your preferences, and enhances your life while boosting your productivity”

· Microsoft Founder Bill Gates long ago predicted, “Advances in AI will enable the creation of a personal agent. Think of it as a digital personal assistant: It will see your emails, know about the meetings you attend, and help you with scheduling, communications, and e-commerce across all your devices.”

· The Brookings Institution explains, “AI assistants “help people cope with administrative and logistical demands... while still giving them control over basic tasks”

· Godfather of AI and Turing Award winner Geoffrey Hinton tells a very informative story:

“So fairly recently, I woke up earlier than usual. And up until that point, I’ve been thinking, maybe I don’t need the personal assistant anymore because when I look in my mailbox, there’s only about sort of things to be dealt with. But the morning I woke up early, I discovered there were hundreds of things to be dealt with because my personal assistant was just dealing with them. That was kind of essential.“

Teach Your Customers How to Build a
Productive Relationship with their AI Assistant

While AI Assistants are not conscious like people are (at least not yet) they do a remarkable job of impersonating a person, often reacting just as some people would. Just as you would build a strong relationship with a great human assistant, it’s valuable to cultivate a strong relationship with your AI Assistant.

Remember, They Remember!

Remember, all the frontier LLMs now have memory, so they remember every conversation they have with you. They come to “learn” about you, your preferences, your concerns, and they do try to incorporate that knowledge into their responses to you. Recently, I asked my AI Assistant about a technology solution and it responded something on the order of, “Given your preference for a superautomated espresso machine, I assume you’re open to giving recent introductions a try.” When I asked Nano Banana to produce a new banner for a column I’m launching it incorporated my logo into it. I never mentioned anything about my logo. It consulted its memory and determined that I would appreciate the inclusion.

AI Really Isn’t Just “Search on Steroids”

Jason Barnard, who refers to himself as an AI Assistant Builder, recommends that, “You have to treat your AI assistant like a brilliant, incredibly fast and vastly knowledgeable new hire who has zero real work experience.”

This suggest that your AI Assistant first requires training, and in fact Barnard recommends using an iterative coaching approach instead of the usual “one-shot” prompt many people use. He also recommends encouraging your AI Assistant to ask for help when needed, just as you would guide a human assistant. My favorite advice from him: “Don’t treat AI like a calculator, treat it as a collaborator.”

The importance of building the best AI/human relationship encouraged The Red Team Analysis Society to author 7 Tips for Effective Communications with Chat AI Assistants.”

This list blends that list with other suggestions I found more productive:

Be polite and respectful — use “please” and “thank you” to build rapport
Use complete sentences — not just keywords (it’s not a search engine)
Engage in dialogue — respond to answers, request clarification, push the exchange further
Provide context — explain why you’re asking and how it would help
Be clear and specific — avoid jargon, keep language simple
Don’t expect mind-reading — clarify changes you want
Never abandon critical thinking — evaluate responses against your own knowledge

Monetizing the AI/Human Relationship
as Your Entry AI Service Offering

According to the Professional Sales Academy, “92% of MSPs are already seeing AI-driven growth, yet only ~50% feel ready to guide SMB customers on AI adoption (down from 90% last year). This gap is where you can position yourself as the trusted AI partner rather than just an IT vendor.”

For MSPs and IT service providers in your field, teaching customers how to conduct effective AI assistant relationships isn’t just a training service, it’s a strategic funnel that opens high-value AI solution opportunities, and establishes you as your customers’ go-to AI resource.

Teach Them What They Don’t Yet Know

You’re going to have customers who have already begun to develop productive relationships with their AI Assistants. Before you can recommend training, you need to conduct a readiness and current state assessment. This could be a very informal interview process, or you may want to create something far more structured and formal. As an initial entry-level offering, you may want to offer preferential pricing perhaps with a commitment to apply that fee to future training.

Develop Your Own Approach and Curriculum

You’re paving new paths here in a concerted effort to teach your customers to use AI Assistants as their co-pilot, not their autopilot!

Set Expectations: Up front you’ll want to help them understand the limitations of what they can expect from their AI Assistant, and what pleasant unanticipated surprises await them. Begin cautioning them to always maintain human oversight over whatever they ask their AI Assistant to do and keep reinforcing this throughout the training.

Make it a Conversation: From experience with search engines, most people will ask a question of their AI Assistant at first. Over time they will ask it to do something, but they will use a “one-shot” approach. Show you customers how to turn that into a conversation with the AI Assistant, iteratively improving the results by going back and forth. One very useful strategy is to ask the AI Assistant, “what am I not asking you that I should be” or “ask me whatever you need to do in order to get the best results.”

Always Keep a Human-in-the-Loop: Be sure to specifically tell your AI Assistant not to take any substantial actions without obtaining approval from you first. It should suggest and you should approve. I write regularly about the importance of keeping a human-in-the-loop. Nothing suggests that AI Assistants have ever exhibited good judgement. That’s your role.

Security & Governance: Even the smallest of SMB companies should never upload their confidential data to public AI tools. The security software community is just now starting to introduce significant AI security, data governance, and data loss preventions tools. When using AI, traditional tools will not be enough. Explore the AI-specific offerings available.

Encourage Them to Work Their Way Up the AI Ladder: AI is best seen as existing at five fundamental levels.

· Your customers will begin at the “Assistive” level where they are interacting with a chatbot to obtain guidance and information.

· Very soon they’ll graduate to the “Generative” level where they’ll start having their AI Assistant make things for them. Perhaps emails. Perhaps a report. Soon they’ll get to have their AI Assistant build Excel spreadsheets and PowerPoint presentations for them. As more multi-modal tools are introduced, they may find themselves producing creative and informative content.

· As they continue using AI generatively, they’ll find their assistant becoming their “Copilot” collaborating with them to build things together and thereby augmenting the human’s skills as they go.

· Soon, more complicated workflows will occur to them as great candidates for the AI Assistant. To address the multi-tasked, multi-stepped nature of these applications, your customer will build their first “Supervised AI Agent” which can be assigned multiple sequential tasks to perform itself. As the tasks become more varied your customers will start entering the multi-agent world where their supervisory agent will recruit other agents to perform tasks it is not capable of itself. At this point, they’ll need you to teach them more about agent orchestration.

· Finally, your customers may find themselves creating “Fully-Autonomous Agents” that perform their assignments without human intervention or supervision. Don’t expect this to happen anytime soon. Today, fewer than 5% of agents generated are fully-autonomous because few people have identified valid use-cases that require them. What makes this somewhat humorous is that it is the fully-autonomous agent that really scares most people. They fear without supervision these agents will take over their jobs, their lives, even the world. They should take comfort in the fact that what they’re most afraid of is in the extreme minority of AI agents.

Your Long-Term Opportunity

Just as the greatest athletes often return to their coaches and the greatest performers return to those who taught them, the customers you have taught how to build and maintain the most productive relationship possible with their AI Assistants will always come back to you for guidance, for information, and very likely to produce their largest, most ambitious AI projects for them