From the Emergency Services edition of October 4, 2026
Spectrum News: Dane County AVA handled 24,000+ non-emergency calls in first month with no major mistakes reported
Dane County Public Safety Communications (Wisconsin) / Spectrum News 1 · County PSAP / emergency communications · United States (Wisconsin)
- Publisher
- Spectrum News 1 Wisconsin
- Original publication
- 2026-09-28
- Source retrieved
- 2026-10-05
- Event date
- 2026-09-28
What happened
Spectrum News 1 reports (Sept 28, 2026) that Dane County's AVA voice assistant handled more than 24,000 non-emergency calls in roughly its first month. The deputy director said there were no major mistakes and attributed misses to how the county programmed the system. Staff rotate monitoring AVA so a communicator can call back if an emergency is missed; callers can ask for a human. Eleven months of the trial remain.
Why it matters
Direct 911-center evidence that conversational AI on non-emergency lines is being run as a staffed, monitored trial, with emergency detection on the non-emergency line called out as a key function.
Evidence and measured results
Local TV reporting quoting a communicator and the deputy director; the 24,000-call count and 'no major mistakes' are agency statements with no published error log, missed-emergency count, or 911 answer-time baseline.
Limitations and uncertainty
One month of operation; metrics are agency-reported; no independent evaluation; 'mistakes' defined informally by the agency.
Put this evidence to work
Lighthouse Advisory interpretation, grounded in this source. Enriched 2026-10-04; this does not change the original publication date. Each role follows the same checklist so takeaways compare across sources.
Sales
Role takeaway
- Customer problem
- Non-emergency calls tie up call takers who also answer 911.
- Stakeholders
- PSAP director/deputy director
- Communicators/dispatch staff
- County board
- Callers needing human access
- Discovery questions
- What share of your lines are non-emergency?
- Who monitors the AI and how often?
- How do you define and count a mistake?
- What happens when an emergency is missed?
- Value hypothesis
- A monitored voice agent may absorb routine non-emergency volume if emergency forwarding and call-back are reliable; value must be shown in recovered dispatcher time.
- Potential engagement
- Trial governance package: error taxonomy, monitoring SOP, and midpoint public report template for a 12-month non-emergency AI trial.
- Unsupported claims
- Faster 911 answer times (not measured)
- Zero missed emergencies
- Staff reductions
Pre-sales engineering
Role takeaway
- Fit
- Non-emergency voice front door with emergency keyword detection and human forwarding.
- Architecture and integration
- Non-emergency trunk to AVA; forward to communicator on emergency cues or caller request; monitoring seat with call-back.
- Prerequisites
- Call-type and keyword rule set
- Human forward and opt-out routing
- Monitoring console and call logs
- Test-call library
- Constraints
- Agency-reported results only
- One month of data
- Keyword detection may miss indirect emergency cues
- Security
- Confirm recording retention, transcript storage, and vendor data-use terms; not described in the article.
- Proof of value
- Missed-escalation rate on seeded test calls
- Share of forwarded calls that were true emergencies
- Dispatcher minutes recovered per week
Delivery
Role takeaway
- Work
- Configure routes and emergency triggers, staff a monitoring rotation, publish caller opt-out, and run a fixed-term trial.
- Dependencies
- Vendor configuration support
- Telephony/trunk changes
- Staff scheduling
- County board oversight
- Ownership
- Public Safety Communications operations owns monitoring, call-backs, and configuration decisions.
- Skills and adoption
- Train communicators to review AI calls, flag misconfigurations, and coach callers to speak naturally.
- Governance checkpoints
- Go-live sign-off on emergency triggers
- Weekly miss review
- Midpoint public report
- Renewal decision at trial end
- Acceptance criteria
- Every suspected missed emergency reviewed within one shift
- Opt-out to human honored on 100% of requests
- Documented error taxonomy in place before go-live
- Risks
- Indirect emergency language not recognized
- Monitoring fatigue
- Over-reliance on agency-defined success
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?
Voice agent on the non-emergency trunk with keyword/warning-sign detection that forwards to a human dispatcher, an opt-out-to-human path, and a staffed monitoring seat with call-back capability.
Governance
Who approves, reviews and stays accountable for outcomes?
A 12-month trial with early public reporting is a reasonable governance model, but 'no major mistakes' needs a defined error taxonomy and published missed-escalation counts to be auditable.
Security and privacy
What data, permissions and controls need testing?
Article does not describe recording retention, vendor data use, or how transcripts are stored; agencies should require these terms before go-live.
Accessibility and workforce
Who is affected, and what skills or accommodations follow?
Staff take turns monitoring AVA, which shifts some dispatcher time to oversight; callers may opt out to a human, which matters for callers who struggle with voice agents.
Procurement
What should contracts, pricing and exit terms secure?
Trial-period structure (11 months remaining) gives the county a decision point; buyers should tie renewal to measured escalation accuracy rather than call counts.
Operating model
Which teams own the service once it runs?
Human-monitored AI front door on the non-emergency line, with misses treated as configuration problems the center owns.
Publication history
- 2026-10-04Emergency Services · Issue 285 resources
Stable resource ID: spectrum-dane-county-ava-first-month-20260928