AI was supposed to make service work lighter. For a lot of MSPs, it made the day heavier.
Kristof Orts, Head of Product and Design, at Thread, has watched that gap open up from the inside. He grew up tinkering on his family’s Macintosh II and fighting his parents for the phone line during the dial-up years. He has spent his career chasing the future technology kept promising. In the movies, that future looks like Tom Cruise swiping through glowing screens with everything at his fingertips.
“This is the future they promised,” Kristof says. “But the reality is more like this. It’s a lot of tools, notifications, channels, dashboards, approvals.”
AI makes service desks feel busier when it adds work to supervise rather than removing work outright. Every new tool brings its own alerts, approvals, and output that someone has to check.
Before AI, technicians typed time entry notes, filled in timesheets, and set the category on every ticket. AI took some of that off their plates. Then it handed them a new job: managing the AI.
“Now you also have to manage your agents,” Kristof says. “You have to track your AI, make sure that they’re not going off script.”
The numbers back him up. 38% of workers say they feel overwhelmed [source]. Time spent on email and messaging keeps climbing. Weekend work keeps creeping in. Cleaning up AI slop has become a task of its own.
That’s the trap. MSPs bolt AI onto a service model that was already fragmented, and the fragmentation gets one more layer.
The work around the work is everything a technician does besides fixing the client’s problem. That covers routing, chasing responses, merging duplicates, writing notes, scheduling, and checking what the AI just did.
None of it shows up as a line item. It hides in a minute here and five minutes there, across every ticket, every day. It’s also the part of the job nobody signed up for.
Thread was built around attacking that layer. When Kristof joined five years ago, the company had four people and a different name. The founding belief was simple: there had to be a better way than managing tickets all day.
The first AI feature Thread shipped, three years ago, went after one small, painful task. Techs hated writing time entry notes. So Thread built AI that read the conversation and wrote the note for them.
“What if AI can just do this all for you?” Kristof says of the thinking behind it.
That question is still the right test for any AI tool on your service desk. Does a piece of busy work disappear, or does it just move somewhere else?
“That’s been our focus,” Kristof says. “To help you manage the chaos and give you back that sense of control.”
Measure the ROI of AI agents as time saved minus the time spent setting them up and keeping them on track. An agent that saves 10 hours a week but takes 8 hours to babysit is barely worth having.
“The ROI from AI is not just time saved,” Kristof says. “You have to also take into account the setup and the maintenance.”
That’s why trust comes before scale. Teams won’t hand off work to an agent they can’t predict. Setup has to build confidence before anything goes live.
Setup measured in minutes, plus maintenance you can actually see, is what keeps the ROI math positive.
AI agents can handle the repeatable coordination work between a request and a resolution. That includes QA before closure, duplicate cleanup, dispatch, and noise filtering, and MSPs are already running all of it.
With Agent Builder, a team describes the job once in plain language, attaches it to a flow, and the agent runs on every ticket. “I want you to merge any duplicate tickets” is a complete set of instructions. Templates get a working agent live in a few minutes.
Here’s what Thread partners have already built:
The same pattern shows up at scale. Since its beta launch, Thread’s auto-dispatch has assigned more than 35,000 threads using scoring, workload limits, and work schedules. Super Magic has handled over a million conversations, more than 500,000 of them in the last 30 days.
Each of those numbers stands for a pile of small tasks that no human had to touch.
The future of MSP service delivery has fewer queues and more conversations. AI agents resolve routine Tier 1 work, every request starts in one place, and technicians spend their time on the actual problem.
Thread’s bets for 2027 point straight at that:
For MSPs, that means less busy work and fewer dropped requests. Technicians get guidance in the moment. Teams can add clients without hiring at the same pace.
“We believe the future of service is not less human,” Kristof says. “It’s really about less busy work. It’s really about having more conversations, not queues. It’s really also about working on the issue, not on the ticket.”
Start with the one piece of busy work your technicians complain about most. Build an agent for that single task, test it on real tickets, and measure the hours it gives back against the time it takes to run.
Then do the next one. That’s how Thread started with time entries, and it’s how the MSPs pulling ahead are cutting the work around the work, one task at a time.
See the agents MSPs are already running and what they’d take off your team’s plate. Book a demo.
Watch the full AISU session presented by Kristof here.