Fewer agents, better agents. That’s the real deal AI offers your service desk.

When automation takes the password resets, the access requests and the “how do I” questions, something uncomfortable happens: the tickets that are left are all hard.

Most organizations plan for the first half of that sentence. Fewer tickets, a smaller team, savings in the budget. Very few plan for the second half.

I’ve led Level 1 and Level 2 teams through this shift, including a global organization of about 38 staff plus a 20-person MSP. As automation and self-service took ticket volume from 60,000 to 42,000, the work that still reached people was more complex, more cross-functional and more visible to executives.

Here’s how I’d restructure Tier 1 and Tier 2 now:

1. Collapse the tiers where you can. If AI handles true Tier 1, your people need Tier 1.5 skills: diagnosis, not scripts.
2. Hire for judgment and communication. The remaining tickets come with frustrated users, unclear ownership and several teams involved. Technical skill is the minimum.
3. Give agents a path, not a pink slip. The best people to tune the AI, write the knowledge and own escalations are the ones who used to work the queue.
4. Change what you measure. Tickets closed per agent stops meaning much. Measure problems prevented, repeat issues removed and customer satisfaction.

AI doesn’t make the service desk smaller and simpler. It makes it smaller and sharper. Plan for the sharp part.

Next in the series: who’s watching the bot? Governance and major incidents when AI is in the loop.

IT leaders who’ve been through this: did you reskill your Tier 1 team or replace it, and what did you learn?