The requester emails askIT@yorku.ca, or the message is forwarded there
01
Halo Auto Triage
SHORTCUT TO SERVICE
Bypassing our biggest bottleneck so solutions start sooner
HIGHEST-VOLUME SERVICE CATEGORIES IN SCOPE
PRODUCTION MONITORING · AUG 19–28
FINAL READINESS GATE · AUG 30
ACTIVE ROUTING · AUG 31
Ticket submitted / forwarded to askIT
Assigned to Auto Triage
Analyzed by Auto Triage
In scope and classified with high confidence?
YESAuto-route
Yes + Duo request missing infoAuto-route + request info
NOManual triage
Service delivered by assigned team
02
01
Request emailed
02
Ticket created
03
Agent reads
04
Agent routes
05
Service begins
Current process / problem
Service waits on triage
Valuable time is spent routing rather than resolving. Before service teams can help, a Client Services agent must read, classify, and route the request
Halo records the new request
A triage agent identifies the intent
The agent selects the category and team
Often days later, the assigned team begins resolving the request
IntakeManual routing workDelivery
03
01
02
03
04
Current risks
Delayed service
Every minute spent moving a ticket delays diagnosis, support, or resolution
Peak times magnify all four risks
Queue delay
Requests wait for a routing decision before reaching the people who can act
Variation
Manual classification can vary from person to person
Back-and-forth
Incomplete Duo requests create a repeat-contact loop and potentially hours of wait
Capacity drain
Skilled agents spend scarce peak capacity moving routine work between queues
04
01
02
03
04
New value
Faster service
Every supported ticket classified with high confidence bypasses manual triage, returning time to service delivery
Faster routing + fewer follow-ups = more capacity for higher-value work
Direct routing
Supported tickets classified with high confidence move directly to the team that can act, so service starts sooner
Consistent decisions
Consistent logic, interpretation, and categorization are applied to every eligible request
Automated follow-up
Incomplete Duo requests receive an immediate request for missing information, reducing a wait that can currently last hours and creating an estimated 45% efficiency gain
Capacity returned
Confidently routed in-scope tickets require no manual classification, freeing skilled agents for higher-value work
05
New process
Automated efficiency
Every incoming AskIT ticket is assessed. Only in-scope, high confidence tickets are auto-routed
01Ticket created in Halo
02Assigned to Auto Triage
03Intent matched to catalogue
04Scope + confidence checked
Direct to resolvers
ClassifyApply the supported service category
AssignSelect the responsible processing team
DeliverThe team reviews, responds, and resolves
No Service Desk handoff required✓
Instant follow-up
ValidateConfirm the defined Duo scenario applies
RespondHalo sends the approved request for information
ResumeService Desk continues when the user replies
One avoidable waiting period is removed✓
Trusted fallback
StopAutomation does not force a classification
ReturnReturn the ticket to the CS queue
TriageA Service Desk agent triages manually
Human judgement remains our fallback↩
06
New risks / New controls
Mitigating risk
The biggest risks remain the same whether a ticket is triaged automatically or manually. New risks have specific mitigations in place
Guardrail
Auto-route only with high confidence
- Launch scope is limited to categories with sufficient training data and demonstrated classification confidence
- Receiving teams can follow current Departmental Triage best practices to mitigate misrouted tickets
- Recurring patterns are reported and used to improve the model. Auto-routing is intended to reduce misrouting over time
Guardrail
Only hunt the white whale: Duo 2FA 🐋
- Send an automatic response only to Duo requests submitted from a York University email address
- Respond only when a Duo request is missing required information
- Halo sends approved text based on an AI trigger, not AI-generated free-form advice
- If the follow-up doesn't apply, a Client Services agent takes over and continues as usual
Guardrail
Dashboard for rapid response
- Stalled tickets appear on the Otto Excellence dashboard
- The UIT Service Desk follows a simple recovery process
Guardrail
Better every day
- During two weeks of silent production monitoring, we compare AI and agent decisions
- We only go live when monitoring meets the agreed readiness criteria
- Low-confidence and out-of-scope tickets stay in the human process
- Every day, we gain more and better data to fine-tune the solution
07
Now → Aug 19
Aug 19–28
Aug 30
Aug 31
Fall peak
What happens when
Rollout timeline
We test the solution in QA and PROD and set readiness gates before it updates or routes a ticket
Build, test, prepare
Complete technical validation, UAT, guidance, frontline training, and operational readiness
Monitoring phase begins
Analyze live production tickets and record decisions without updating, assigning, or routing them
Readiness Gate
Assess accuracy, exceptions, operations, and people readiness before active routing
Active routing begins
Supported tickets classified with high confidence begin to be routed automatically. All others follow normal triage
August 31 is a target contingent on a successful monitoring period
Hypercare and refinement
Monitor metrics and stalled tickets, recover exceptions, and continue fine-tuning based on the results
08
Why right now?
Highest volume = highest value
Start with a small scope
Only the most frequently used categories are eligible
Require high confidence
Uncertain decisions stay with the Service Desk
Humans remain the heart
People remain responsible for exceptions, judgement and service quality
Observe before acting
Two weeks of production data inform the decision
09
Impact matrix
Who feels the change?
Requesters
Low impact
What changes
- Some requests reach the people who can help sooner
- Incomplete Duo requests may receive an automatic reply requesting missing information
What stays
- They continue to use the same service channels
- The same support staff will resolve their requests
Required action
- Respond to automatic follow-ups when requested
IT agents
Low impact
What changes
-
Some tickets will be automatically routed to the following queues:
- CTS
- Classroom Operations
- ED&I
- Learning Technology Services
- UIT Service Desk
- Telecommunications
- Tickets may reach all queues faster than before, whether routed automatically or manually
What stays
- Current review, response, escalation, categorization, and resolution practices remain in place
Required action
- Monitor ticket queues early and often
- Follow Halo best practices, especially for handling misrouted tickets and verifying categorization before resolution
-
Notify OpEx about persistent routing errors
- Raise an Incident to OpEx: Application > HaloITSM > Triage Agent Issue
UIT Service Desk
Medium Impact
What changes
- Less manual routing for confidently assessed in-scope ticket
- The Service Desk monitors and recovers stalled tickets
What stays
- The Service Desk continues to triage low-confidence, out-of-scope, incomplete, and exceptional requests
Required action
- Monitor the queue and Otto Excellence Dashboard
- Avoid interrupting active Otto Excellence processing
- Recover stalled tickets
-
Notify OpEx about persistent routing or stalling errors
- Raise an Incident to OpEx: Application > HaloITSM > Triage Agent Issue
IT leaders and managers
Low impact
What changes
- Leaders receive clearer operational evidence about routing performance, exceptions, and readiness
What stays
- Service ownership, escalation accountability, and human responsibility remain with people
What we need
- Reinforce expectations
- Confirm team readiness
- Support evidence-based tuning during peak
10
01 · RECOGNIZE
02 · MONITOR
03 · RECOVER
04 · REPORT
UIT Service Desk training
Prepare the Service Desk to...
With capacity freed from manual routing, the Service Desk will monitor Auto Triage and remediate issues
Understand the process
Recognize when Otto Excellence is actively processing and avoid interrupting Auto Triage
Watch for issues
Check the IT Client Services queue and Otto Excellence Dashboard often
Route stalled tickets
Confirm that a ticket is stalled, then triage it as usual
Turn patterns into improvement
Record and report recurring errors, unusual trends, and repeated stalls, then notify Operational Excellence
11
The stage is set
How it plays out
Major value with minimal disruption
Aug 11
All IT staff
Change Advisory 1
First change advisory notifying IT staff of the Halo change and directing them to Auto Triage resources for more information
Aug 19
OpEx
Begin monitoring
The monitoring phase begins, but nothing changes for agents or requesters
Aug 26
All IT staff
Change Advisory 2
Second change advisory notifying IT staff of the planned move to active routing
Aug 19–30
IT agents
Prepare agents
Send IT agent communications with links to the IT agent guide. Begin Service Desk-specific training
Aug 30
IT leadership
Confirm readiness
Analyze and share the monitoring-phase data and confirm readiness for active routing
Aug 31
All IT staff
Launch
Activate and monitor auto-routing
12
01
02
03
Champion the change
Prepare our people
Otto shuffles the tickets
We shift the culture
Raise the banner
Herald Auto Triage and reinforce the message that it removes waiting and protects scarce capacity. It brings service closer to our community; it doesn't remove the people who deliver it
Rally your people
Confirm your teams understand the coming change and their roles in it. Direct them to the communications and resources being shared. I’ll also send everyone a one-page reference after this meeting
Fortify our foundation
This is the foundation of service automation. We need to hold the line through the early rollout, learn from every result, and build the operational discipline needed to scale automated routing and responses
SEQUENDA VIA OUR PATH FORWARD