The one-FTE math: how AI Triage actually recovers 22 hours a week
A breakdown of where the dispatcher's hours go — and how AI Triage gives most of them back without compromising the client experience.
"AI saves you a headcount" is a marketing claim that means almost nothing without showing the math. Here's the actual math we see across MSPs in their first 90 days running AI Triage.
Where the dispatcher's hours actually go
The unspoken truth about MSP service desks: a senior dispatcher spends 60–70% of their week on tasks that don't require senior judgment. We measured a 6-tech MSP for two weeks before rollout. Their lead dispatcher's time looked like this:
- 22 hours/week reading inbound tickets and assigning category, priority, SLA, and a first-pass tech.
- 9 hours/week chasing missing context — calling the requester back, tagging contracts, looking up CIs.
- 6 hours/week updating clients with status while a tech worked the ticket.
- 5 hours/week in the dispatch huddle.
- 3 hours/week on legitimate dispatch judgment — the calls about who is the right escalation path, what to deprioritize, what to push to next sprint.
That's 45 hours of work in a 40-hour week. The 5-hour overflow shows up as overtime, missed lunches, or things falling through the cracks.
What AI Triage actually replaces
Of those 22 hours of triage time, AI Triage classifies more than 95% of tickets correctly inside 3 seconds — including category, priority, contract SLA, suggested tech, and known-issue dedup. The dispatcher doesn't disappear. They review the AI's classification, override the 4–5% it gets wrong, and use the recovered time for the harder dispatch judgment that actually requires them.
"The dispatcher didn't get fired. They got their job back. The bottom of the funnel — the routing busywork — got automated. The top of the funnel — the human judgment — got more attention."
What it doesn't replace
AI Triage doesn't:
- Talk to clients. That stays human (or gets handed to AI Voice on the inbound channels).
- Make escalation calls about contract politics.
- Decide whether to invoke change-management for a P1 fix.
- Handle the moment a tech is overloaded and you need to rebalance the queue mid-day.
If you replaced your dispatcher with the model, things would break. The point is to replace the routing busywork, not the human.
The one-FTE math
Three numbers in particular drive the FTE-equivalent recovery:
- 22 hours of triage → AI handles ~21 of them. Net 21 hours back.
- 9 hours of context-chasing → AI Voice + identity verification at ingest captures most of this upfront. Net ~6 hours back.
- 6 hours of client status updates → automation bots and outbound notifications cover the routine "still working on it" updates. Net ~4 hours back.
Total: ~31 hours of weekly capacity recovered for one dispatcher. That's not a full 40-hour FTE on paper, but it's enough to grow the book by 30–50% without adding a senior dispatcher hire — which is the actual decision the math is supporting.
The honest framing
AI Triage doesn't save you a person. It saves you the next person you were about to hire. That's the right comparison, and it's the comparison that makes the ROI math work.
The other thing the dispatcher gets back: their attention. A senior dispatcher who's spending 22 hours a week on rote classification is bored, distracted, and overbooked. The same dispatcher with 22 hours back finds problems before they become tickets, builds better runbooks, and trains the techs more.
That second-order effect is harder to quantify. It's also bigger than the hours.
Ship a verified service desk in 30 days.
Book a 30-minute call with a solutions engineer who came out of an MSP service desk. Bring your stack. We'll model the impact with your numbers.