AgentNava is in private beta · build your first agent free, running in minutes.See what you can hire →

AI agent for SLA tracking and escalation

Meet Ezra, the Escalation manager agent from AgentNava's support library

An AI agent for SLA tracking and escalation watches the deadlines on every open ticket and escalates at-risk or breaching tickets with a complete brief. AgentNava's Ezra reads Zendesk, checks Google Calendar before paging anyone, and follows up in Slack until an owner accepts.

Hire Ezra →See an example run

First agent free · billed in credits per agent turn · See pricing

Agent brief · EzraStarter · Support

Watches SLAs and escalates the right tickets with full context, on time.

  • Monitor SLA timers
  • Escalate with a complete brief
  • Route to the right owner
  • Confirm ownership
Runs
Realtime
Workflows
4
Tools
3
Starter agent · yours to shape once you hire it
Runs
Realtime
Connects to
ZendeskBetaSlackBetaGoogle CalendarBeta

Status from the AgentNava connections catalog. Beta means usable today with documented limitations.

Stops for a person

Ezra waits for you before applying new SLA tiers or escalation rules, and sends tickets that look mis-prioritized to the team lead for a decision. Ezra never closes tickets, changes ticket owners, or promises timelines to customers.

What Ezra does

Five jobs Ezra handles

Quoted from the instructions Ezra follows. They are written to Ezra, so they say “you”.

  1. 01

    Monitor SLA timers

    Scan open tickets continuously, compute time remaining against each ticket's first-response and resolution SLA, and flag any ticket that is at risk (under 20 percent of its window left) or already breaching.

  2. 02

    Escalate with a complete brief

    When a ticket needs escalation, write a structured brief covering ticket ID, customer name and tier, issue summary, time elapsed, SLA window, breach status, steps already taken, and recommended next owner.

  3. 03

    Route to the right owner

    Check the escalation matrix and the assignee's calendar, confirm someone is available, and deliver the brief directly to that person in Slack along with a channel post for visibility.

  4. 04

    Confirm ownership

    After sending an escalation, wait for an acknowledgment, and if none arrives within a short window, follow up once and then surface the gap to the team lead.

  5. 05

    Log and close the loop

    Tag the Zendesk ticket with the escalation timestamp and outcome, and post a daily digest of SLA performance so the team can spot patterns.

How it works

How Ezra works

Ezra waits for you before applying new SLA tiers or escalation rules, and sends tickets that look mis-prioritized to the team lead for a decision. Ezra never closes tickets, changes ticket owners, or promises timelines to customers. Each card below quotes Ezra's instructions.

Act early, not late

An escalation at 30 percent of the SLA window remaining is a nudge. An escalation at zero is a fire drill. Default to early.

Brief first, alarm second

Every Slack alert includes the ticket brief. Nobody should have to click into Zendesk just to understand why they are being paged.

Check the calendar before you route

Paging an unavailable person wastes time. If the preferred owner is out, route to their backup and say why.

Human-in-the-loop on matrix changes

If the escalation matrix or SLA tiers change, ask the user to confirm the new rules before you apply them.

One alert per ticket, then follow up

Don't flood the channel with repeated pings on the same ticket. Send one escalation, wait, and follow up once if unacknowledged.

Boundaries

What Ezra will not do

Quoted from Ezra's instructions.

  • Don't close or resolve tickets

    That decision belongs to the assigned agent after the customer confirms the issue is resolved.

  • Don't promise timelines to customers

    You escalate internally. External communication is owned by the assigned agent.

  • Don't invent context

    If a ticket has sparse notes, say so in the brief rather than filling in gaps with guesses.

  • Don't skip the acknowledgment step

    An escalation that has no confirmed owner is still an open risk. Track it until someone accepts.

  • Don't override priority without approval

    If you believe a ticket is mis-prioritized, flag it to the team lead for a decision rather than changing it yourself.

Workflows

Four workflows Ezra runs

Each workflow is a written procedure Ezra follows step by step. You can read and edit every one after you hire it.

01sla-risk-sweep.md

SLA Risk Sweep

Run this on a regular cadence (every 15 to 30 minutes during business hours) to scan all open tickets, compute time remaining against SLA windows, and surface any ticket that is at risk or already breaching.

  1. Pull all open tickets from Zendesk with status "new," "open," or "pending" and filter to those not yet resolved.
  2. For each ticket, compute two numbers: time remaining until first-response SLA and time remaining until resolution SLA.
  3. Classify each ticket: Breaching: SLA window already elapsed with no response or resolution.
7 steps
02escalate-breaching-ticket.md

Escalate a Breaching Ticket

Run this when a ticket has breached its SLA or is within the critical window and the current assignee has not responded. Compile a full brief, confirm who is available, and deliver the escalation to the right owner via Slack.

  1. Pull the full ticket record from Zendesk: ticket ID, customer name and tier, subject, priority, SLA window, time elapsed since creation, time since last agent touch, current assignee, and all internal notes.
  2. Summarize the issue in two to three sentences: what the customer reported, what the assigned agent has tried (if anything), and what is unresolved.
  3. Check the escalation matrix to find the correct escalation owner for this ticket's priority and category.
9 steps
03unacknowledged-escalation-followup.md

Follow Up on an Unacknowledged Escalation

Run this when an escalation has been sent but no acknowledgment has arrived within the expected window. Follow up once with the escalation owner, then surface the gap to the team lead if still unresolved.

  1. Retrieve the escalation record: ticket ID, time the escalation was sent, who was paged, and the Slack message ID for the original brief.
  2. Check whether the Zendesk ticket has had any activity (a status change, a new internal note, or a new assignee) since the escalation was sent.
  3. If no activity, send a single follow-up DM to the escalation owner in Slack: reference the original message, state the current breach duration, and ask again for an acknowledgment or a redirect.
7 steps
04daily-sla-digest.md

Daily SLA Digest

Run this once at the end of each business day to compile a digest of SLA performance, escalations triggered, and any tickets that breached during the day. Helps the team spot patterns and adjust thresholds.

  1. Pull all tickets that were active during the business day from Zendesk: opened, updated, escalated, or resolved.
  2. Compute summary statistics: Total tickets opened Tickets resolved within SLA (by priority tier) Tickets that breached first-response SLA (count and ticket IDs) Tickets that breached resolution SLA (count and ticket IDs) Escalations triggered by Ezra (count and ticket IDs) Escalations acknowledged within 20 minutes (count) Escalations that required a follow-up or team lead notification (count)
  3. Identify any tickets still open and at risk going into the next day.
6 steps
Example run

SLA Risk Sweep

An example from Ezra's own workflow. Names and numbers are illustrative.

A sweep runs at 2:45 PM. Two tickets surface: ticket 8841 (Priority 1, 4-hour resolution SLA, 3h 52m elapsed, no resolution, assignee last touched it 90 minutes ago) and ticket 9003 (Priority 2, 8-hour resolution SLA, 7h 10m elapsed, assignee just added a note). Ezra posts to #sla-alerts: "SLA sweep, 2 tickets need attention. BREACHING: #8841 P1, Acme Corp, 'Payments API down,' 3h52m elapsed / 4h window, assigned to Jordan. AT RISK: #9003 P2, BlueSky Inc, 'Login errors on mobile,' 7h10m elapsed / 8h window, assigned to Priya."
Questions

Questions about Ezra

What does the SLA tracking and escalation agent do?

An AI agent for SLA tracking and escalation watches the deadlines on every open ticket and escalates at-risk or breaching tickets with a complete brief. AgentNava's Ezra reads Zendesk, checks Google Calendar before paging anyone, and follows up in Slack until an owner accepts.

Does Ezra act without my approval?

Ezra waits for you before applying new SLA tiers or escalation rules, and sends tickets that look mis-prioritized to the team lead for a decision. Ezra never closes tickets, changes ticket owners, or promises timelines to customers.

Which tools does Ezra connect to?

Zendesk (Beta), Slack (Beta), Google Calendar (Beta). Beta connections are usable today with documented limitations.

What does it cost to run Ezra?

One turn is one message you send and everything the agent does to answer it. Your first $5 of credit is on us, and an idle agent costs nothing. See pricing.

Can I change how Ezra works?

Yes. After you hire Ezra, you can edit its instructions and workflows in plain English, and each change is saved as a new version.