Maya is a tier 2 tech at Acme MSP. She's been doing this for three years, and she's good at it. But every morning she walks in to the same scenario: twenty new tickets sitting untriaged from overnight, leaving her to start her day by scrolling the Autotask queue.

Conduit's ticket-triage automation changes that.

Let's walk through four moments in Maya's day, before and after Conduit.

7:55 a.m.

Before Conduit. Maya sits down with coffee and opens Autotask. Twenty new tickets have come in since last night, all sitting at status "New," and none are prioritized. She starts reading. Some are obvious (a password-reset request, a printer issue). Some aren't (a vague "email is weird" that could be nothing or the first sign of a phishing incident spreading through a client). She reads each one, assigns the priorities in her head, and updates the records. It takes her 20 to 30 minutes before she's actually working a ticket.

After Conduit. Maya sits down with coffee and opens Slack. A message from the triage agent is waiting: twenty tickets triaged since her shift ended. Two Critical, three High, five Medium, and five Low. Every ticket already has priority set in Autotask and status moved to In Progress. The agent even flagged one it couldn't confidently classify and posted a "needs a human" note. Maya opens Autotask and starts on the Critical ones. Total elapsed time before real work starts? About 2 minutes.

10:30 a.m.

**Before. **Maya is deep in a firewall config when a ticket comes in from a client who always writes vague subject lines. It sits at New, and nobody notices for 40 minutes. When Maya finally sees it (or when the client calls to follow up), it turns out to be an urgent MFA lockout that’s affecting the client's CFO. Maya apologizes for the delay.

**After. **The agent runs on the hour. When a ticket comes in at 10:22, it's picked up by the 11:00 run, classified as High, and Slack pings the channel. A dispatcher sees it in the Slack feed, and Maya gets tagged in before the client feels the need to follow up. The lockout gets resolved cleanly and quickly.

A note on that hourly cadence: Claude's scheduled routines have a one-hour minimum, so triage runs on the hour. In practice, that's the right cadence for most MSPs. Ticket triage in batches, once an hour, is more workable than being interrupted every 90 seconds by a bot that thinks every ticket is Critical.

1:00 p.m.

**Before. **Maya comes back from lunch. She scans the queue for anything new since 11:30, sees five more tickets, reads each one, prioritizes, and assigns. It takes about 15 more minutes before she's back to actual work.

**After. **She glances at Slack. The 1:00 p.m. triage summary is already there, with two new tickets, both Medium, and both In Progress. She scrolls the two, confirms the agent's calls, and gets on with her afternoon.

5:30 p.m.

Before. Maya writes a handoff note for the on-call tech: what's open, what's escalated, and what the client called in about at 4:47 p.m. that isn't in the system yet. It takes her 20 minutes of writing, checking her memory against the queue, and remembering the context she was going to include, but forgot.

**After. **The queue is the handoff. Every ticket has a priority, a status, and an In-Progress marker from a triage run. The Slack channel has a chronological trail of everything that came in and got classified. The on-call tech sees the state of the queue at a glance, and Maya's handoff note is two sentences, instead of two pages. She's out the door at 5:35.

What Adding Conduit to Your Shop Changes

The tickets didn't disappear, and the clients didn't stop needing help. But Maya spent her day working tickets, instead of managing the queue.

The Conduit ticket triage agent runs on a schedule, reads new tickets in Autotask, classifies them by priority, moves them to In Progress, and posts a summary to Slack. If it can't confidently classify a ticket, it leaves the ticket alone and flags it for a human. Nothing gets silently mis-triaged, and nothing gets lost.

The full walkthrough (with the actual build prompt) lives in the Conduit docs. First-time setup takes about 20 minutes.

If you’re curious what MSP queue triage automation could look like for your shop, let's talk.