Stop doing it twice: Use AI to finally automate the work you hate – by Rewst
By Rewst | Published on MSPGeek
There’s a principle that shows up early in good engineering: only do something once, because you only have one failure point. If a process has to run a second time by hand, that’s where variance creeps in. A step skipped, a field entered differently, an alert missed because the right person wasn’t available. Engineers who’ve internalized this don’t think of automation as extra work. They think of repeating a manual process as the extra work.
“One of the stronger engineering principles in building software is to only do something once, because you only have one failure point. Engineering-minded people already have that notion: if I have to do something more than once, I’m going to automate it.” — Johan Van Heerden, SVP of Engineering, Rewst
That discipline is good engineering. The Rewst 2026 State of MSP Automation, a report based on a survey of 300 MSP professionals across North America, EMEA, and APAC, revealed that the MSP industry has almost universally arrived at the same conclusion.
97% of MSPs plan to automate more processes this year
95% say automation is no longer optional
The conviction is universal. But conviction alone doesn’t close what the Automation Report calls the Automation Divide: the growing split between MSPs building automation as a strategic capability and those still running on early wins and manual fallbacks. So what’s getting in the way?
The work worth hating
Not every process deserves to be automated, but the ones worth hating usually do. The profile is consistent: high frequency, well-defined outcome, and few edge cases. If you can describe the inputs, the steps, and the expected output, and it happens over and over, that’s your target.
Documentation is a perfect example. Our QA team at Rewst used to manually pull together deployment data, cross-reference code changes, and synthesize a risk profile before every release. Now that data is fed into AI and a risk report is generated automatically. What used to take hours of human effort runs daily without anyone thinking about it.
That’s what the work you hate usually looks like up close: not one catastrophic time sink., It’s a cluster of tasks, five or ten minutes each, repeated constantly, that add up to an end-to-end process and drain the energy of your most capable people on work that doesn’t require their judgment.
“It’s the same as compound interest. Those five-minute tasks tend to be high-frequency tasks, and over six months, five minutes becomes days or weeks that you can save. Every five minutes you reclaim creates more opportunity to save another five minutes.” — Johan Van Heerden
Most MSPs are climbing the same ladder without realizing it
Here’s something from the report that’s worth paying attention to: there’s a clear progression to how MSPs incorporate AI into their automation work, and most are climbing it in stages.
88% of MSPs use AI to write scripts and documentation. This is where almost everyone starts. The complexity is low: you know what each step needs to do, the sequence is clear, and AI is essentially transcribing logic you already understand.
62% use AI to build automations. This is where the thinking gets harder. You’re no longer executing a known sequence. You’re accounting for edge cases, variances in the data, and conditions that branch the logic. AI is helping you think through the structure, not just write the steps.
Only 44% have AI making decisions inside their workflows. This is the level where context becomes everything. You’re asking AI to evaluate incoming data (sentiment, risk, patterns) and either route it or act on it.
Want to dig into these numbers with the people behind the report? Aharon Chernin, Frank Price, and Jennifer Roy (CEO, Nucleus Networks) are breaking down the findings live on May 14. Register for the webinar.
“When you write a script, the complexity is generally fairly low because you know what each step needs to do. The moment you start to automate, you start thinking about edge cases. And on the decision-making side, do you have the right context available? The complexity of the context and the information you give on each of those phases gets higher.” — Johan Van Heerden
Understanding where you sit on that ladder is useful. Most MSPs are at the first or second step, and the next move is clear. But the reason to get past script-writing isn’t just about handling more logic. It’s about what automation gives you that scripting doesn’t: built-in auditing, monitoring, and consistent failure handling. Adding those to a script is a heavy lift, and that’s where the hidden cost kicks in. Before long you’re spending more time on the infrastructure than on the actual automation. That overhead compounds. It’s worth knowing what you’re signing up for before you commit to building from scratch.
AI thinks. Automation acts. You need both.
One of the most common sources of confusion in this space is treating AI and automation as interchangeable. They aren’t, and the distinction matters practically.
AI is good at pattern recognition, synthesis, and planning. Given enough context, it can identify what’s likely wrong in a log file, draft a workflow from a description, or flag a deployment as high-risk before a human would. Our engineering team cut troubleshooting time from four days to roughly an hour by feeding log data and code paths to AI and letting it identify the pattern.
But AI on its own doesn’t give you a repeatable process. It gives you an answer in this instance. Automation executes on that answer the same way every time, across every client. It’s also what keeps your institutional knowledge in the workflow itself, accessible to the whole team, rather than locked inside any single person’s head. Our CEO Aharon Chernin is clear that you need both: AI can help you design, test, and improve a workflow. But the workflow has to exist.
This is also why the expertise problem is stickier than most MSPs expect. The Automation Report found that lack of expertise is the top blocker at 54%, and it doesn’t shrink as MSPs grow. MSPs in the $10M to $25M revenue band cite it at 63%, compared to 50% for smaller shops. Budget, by comparison, came in dead last at 18%.
For a long time, progress on automation depended on a small group of people who knew how to build. Everyone else had ideas and waited. Even when the opportunity was obvious, execution sat with whoever had the right skills. Our CPO, Frank Price, calls this the blank sheet problem, and it’s one of the things AI is starting to solve. More of the team can now describe what they need and get a working starting point without mapping every step in advance. Expertise still matters, but it shows up differently: less time translating intent into structure, more time refining workflows and deciding where automation actually belongs.
There’s also a hidden cost to going it alone with scripts. They don’t come with auditing, monitoring, or consistent failure handling built in. You end up constructing all of that yourself, and before long you’re spending more time on the infrastructure than on the actual automation. That overhead compounds. It’s worth knowing what you’re signing up for before you commit to building from scratch.
What gets built when you stop doing work you hate
There’s a version of the automation conversation that’s entirely about efficiency: save time, reduce errors, handle more tickets with the same headcount. That’s real. Operational efficiency is the most frequently reported outcome at 86%, followed by consistency and standardization at 74%.
But the more interesting outcome is what happens to the people who used to do that work.
“When a team stops doing the work they hate by hand, they become more creative, they become more innovative. They get more opportunity to get closer to the customer and the actual scenarios they need to worry about.” — Johan Van Heerden
You also shift from reactive to preemptive. When your team isn’t triaging the same issue for the hundredth time, it can look at the pattern of issues and ask why it keeps happening. Your technicians can take today’s logs, run them through AI, spot recurring error signatures, and address them before they become tomorrow’s tickets.
This is the Automation Divide showing up in the data:
68% of executives strongly agree that AI + automation together create more value than automation alone
2x+ the rate of Automation-as-a-Service delivery for MSPs with a dedicated automation platform (23% vs. 9%)
90% AI usage rate among MSPs at the highest automation maturity (vs. 26% at the lowest)
The Automation Divide is real. And it’s widening.
The first step is the one you already know.
You don’t have to figure out the full roadmap to start. The place to start is the task you genuinely dread: the one that shows up in the queue every single day. The one where you think “this again” before you even open the ticket.
That’s the one. High frequency, well-defined, probably more straightforward than it feels once you document the process. You already know the inputs. You already know what good looks like. And with AI lowering the barrier to building, the distance between “I should automate this” and “this is automated” has never been smaller.
Stop doing it twice. Automate it once. Then use the time you get back to build the next one.
We surveyed 300 MSPs to find out where the Automation Divide is actually forming. See where you stand.
Get the 2026 State of MSP Automation →
