From the Emergency Services edition of October 22, 2026
Ozaukee Press: county's AI non-emergency line still frustrates some callers six months in, with no estimate of staff hours saved
Ozaukee County Sheriff's Office Communications Center · Emergency communications (911 / dispatch) · Ozaukee County, Wisconsin, United States
- Publisher
- Ozaukee Press
- Original publication
- 2026-10-21
- Source retrieved
- 2026-10-23
What happened
Ozaukee Press reports the sheriff's office AI phone system Ava has answered non-emergency calls since April, earlier than its public introduction, and initially struggled to route calls. A former dispatcher who reported a disabled car hung up after Ava asked for his personal details; the record shows a dispatcher reviewed the call at 1:32 p.m. and a deputy was sent at 1:33 p.m., two and three minutes after the 1:30 p.m. call. 911 calls still go to humans, dispatchers review Ava's calls, staffing will not change, no estimate of hours saved exists, and the cost was about $70,000, largely federal grants.
Why it matters
A U.S. consolidated center serving seven police and seven fire departments shows the caller-experience and measurement gaps that follow AI non-emergency lines after launch.
Evidence and measured results
Local newspaper reporting with quotes from the undersheriff and sheriff, one caller account and call-record times supplied by the agency; no performance data.
Limitations and uncertainty
Single caller anecdote; timing figures come from the agency; Ozaukee's Ava deployment and 17% call growth are already in the archive (WRJN, TMJ4), and this item adds the April start, caller-experience problems, cost and absence of savings data.
Put this evidence to work
Lighthouse Advisory interpretation, grounded in this source. Enriched 2026-10-22; this does not change the original publication date. Each role follows the same checklist so takeaways compare across sources.
Sales
Role takeaway
- Customer problem
- Dispatchers handle over 113,000 non-emergency calls a year while 911 volume rises, and the AI line meant to help has unclear benefits.
- Stakeholders
- Sheriff and undersheriff
- Communications center manager
- Dispatchers
- Fire and police chiefs served
- County board
- Discovery questions
- How many Ava calls are abandoned?
- How quickly do dispatchers review AI-taken calls?
- What hours-saved measure would satisfy the board?
- Value hypothesis
- With better intake design and measurement, an AI non-emergency line could free dispatcher attention without hurting caller experience.
- Potential engagement
- Post-launch measurement and caller-experience review of an AI non-emergency line.
- Unsupported claims
- Any staff hours saved
- Improved caller satisfaction
- Fewer dispatch errors
Pre-sales engineering
Role takeaway
- Fit
- Conversational AI intake and routing for a consolidated center's non-emergency line.
- Architecture and integration
- Vendor voice agent on the non-emergency line, routing to divisions and transferring emergencies to 911, with records reviewed by dispatchers.
- Prerequisites
- Call logging and abandonment tracking
- Intake script control
- Vendor tuning process
- Constraints
- Callers' phone-tree habits
- Bilingual callers
- Small county budget
- Security
- Limit personal data collected on incident reports and set retention rules for AI call records.
- Proof of value
- Abandoned-call rate before and after script changes
- Median time to dispatcher review
- Routing accuracy by division
Delivery
Role takeaway
- Work
- Stabilize and measure an AI non-emergency line after launch.
- Dependencies
- Vendor cooperation
- Public education
- Dispatcher review workflow
- Ownership
- Ozaukee County Sheriff's Office Communications Center.
- Skills and adoption
- Dispatchers review AI-taken calls; the public learns to state the reason for calling.
- Governance checkpoints
- Public notice of AI use
- Quarterly performance review
- Data collection policy check
- Acceptance criteria
- Proposed: published abandoned-call and review-time figures
- Proposed: staff-hours estimate within one quarter
- Risks
- Caller frustration deterring reports
- Unmeasured value
- Over-collection of personal data
Implementation considerations
Lighthouse Advisory interpretation across the operating dimensions a public-sector buyer must settle before this evidence becomes a design. Each note answers the question under its heading for this specific source.
Architecture and integration
What must connect, and where does the AI sit in the workflow?
Ava routes non-emergency calls to divisions, transfers emergencies to 911, supports Spanish, and logs calls for dispatcher review; installed as part of a 2025 dispatch upgrade.
Governance
Who approves, reviews and stays accountable for outcomes?
Human dispatchers review Ava's calls; the agency launched quietly before a public introduction, which raises transparency questions for callers.
Security and privacy
What data, permissions and controls need testing?
The caller objected to being asked for name, job and address on a roadside report; data collection scope and retention are not described.
Accessibility and workforce
Who is affected, and what skills or accommodations follow?
Spanish support is offered; callers unused to conversational systems may struggle; no staffing reductions planned.
Procurement
What should contracts, pricing and exit terms secure?
About $70,000 largely from federal grants; contracts should require performance reporting.
Operating model
Which teams own the service once it runs?
Dispatchers shift to reviewing AI-taken non-emergency calls while 911 stays human.
Publication history
- 2026-10-22Emergency Services · Issue 464 resources
Stable resource ID: ozaukee-press-ava-nonemergency-ai-caller-frustration-no-hours-estimate-20261021