AI is everywhere in the MSP space right now. New tools, new promises, new pressure to move fast or get left behind. But as Thread COO Matt Linn shared in a recent webinar with Steve Taylor of Alternative Payments, the real challenge is not whether MSPs should adopt AI. It is how they do it.
Jumping straight into AI without a plan does not create efficiency. It creates frustration, failed implementations, and missed outcomes. What MSPs actually need is a maturity model. A roadmap that meets them where they are and helps them progress with confidence.
That is exactly what Matt and Steve unpacked in their conversation. How AI fits into the service desk over time, why skipping steps almost always backfires, and how MSPs can use AI to make service more human, not less.
Matt’s perspective is shaped by more than a decade inside an MSP. Before Thread, he spent ten years at Richard Fleischman and Associates, growing from compliance and resiliency work into operations leadership. Along the way, he watched MSPs navigate multiple industry shifts.
First came cloud computing. Then security moved from an afterthought to a core revenue driver. Now AI represents the next major evolution. Not as a shiny add on, but as a foundational capability MSPs must learn to operate with.
Like every shift before it, AI is not something you simply turn on. MSPs have to implement it internally, understand it operationally, and only then begin using it to drive real business outcomes.
Before any AI enters the picture, Matt emphasizes something most maturity models overlook. Step zero.
This is the moment of honest introspection. Is your organization actually ready to adopt AI. Are leadership and frontline teams aligned on why you are doing it. Do you have the capacity, resources, and willingness to change how work gets done.
Without that cultural alignment, both top down and bottom up, AI initiatives stall. Tools do not fail because the technology is bad. They fail because the organization was not ready for the change.
Stage one is where most MSPs start, whether they realize it or not. Tickets are created manually. Priorities are assigned by humans. Routing decisions depend on experience, judgment calls, and internal tribal knowledge.
Even in best case scenarios, Matt notes that manual triage and classification typically max out around fifty five to sixty percent accuracy. Different dispatchers work differently. Taxonomies drift over time. Service catalogs evolve. Every inconsistency compounds.
This stage is not bad. But it does require discipline. Cleaning up ticket types, subtypes, priorities, SLAs, and workflows creates the foundation AI needs to work effectively later.
Stage two is where AI starts doing real work, quietly.
Assistive AI handles repetitive and low value tasks behind the scenes. Categorization, prioritization, routing, time entry, and documentation. Requests still come in through chat, email, or phone. Technicians still deliver the service. But the manual clicks disappear.
The key difference is that AI is not making decisions for humans yet. It is removing friction around them.
Technicians gain what Matt calls superpowers. They spend less time updating tickets and more time solving problems. The experience improves without customers even realizing AI is involved.
Stage three is where trust starts to matter.
Agentic AI can reason, infer, and act. Instead of just assisting, it begins to own specific tasks. Matt recommends starting internally and behind the scenes before putting AI directly in front of customers.
Examples include automatically updating documentation at the close of a support session or keeping knowledge bases current based on real world issues being resolved. These agents deliver immediate value while minimizing risk, which makes them ideal for early experimentation.
From there, MSPs can expand into AI led triage and initial engagement through chat, email, and eventually voice. The goal is not to replace human service. It is to get customers help faster, with less friction and more consistency.
As the workforce skews younger, preferences are shifting. More users want to chat instead of calling or emailing. They want support where they already work, inside Slack, Teams, or desktop tools.
Matt points out that this shift does more than improve convenience. It dramatically increases engagement. MSPs using chat based workflows often see CSAT response rates jump from the teens into the eighty to ninety percent range, simply because feedback is captured in the moment instead of buried in email surveys.
With AI now moving into voice, the same principles apply. Faster resolution. Less waiting. A more human experience powered by automation.
The biggest takeaway from the conversation is simple. You cannot skip maturity stages and expect good outcomes.
Calling yourself stage two does not make it true. AI amplifies what already exists. If your data is messy, your workflows inconsistent, or your culture misaligned, AI simply helps you fail faster.
But when MSPs take a structured approach, starting with clean foundations, layering in assistive AI, and gradually adopting agentic capabilities, the results are transformational. More capacity without more headcount. Better margins. Happier technicians. Better customer experiences.
Or, as Matt puts it, using AI to make service more human.