Customer support · Production

City of Midland cuts missed calls and scales citizen support with ElevenLabs-powered civic concierge 'Jacky'

The problem

Midland managed more than 3,000 inbound calls per day, with peak-time missed connections and longer resolution times. Traditional IVR menus made complex requests harder to complete quickly, and language coverage did not always match the multilingual community's needs.

Workflow diagram · grounded in source
1
Overflow call routed to Jacky
routing
“Overflow calls now route to Jacky for immediate, natural assistance that understands context and speaks with empathy”
2
Jacky handles call with voice AI
ai_action
“a civic concierge powered by ElevenLabs Agents. Overflow calls now route to Jacky for immediate, natural assistance that understands context and speaks with empathy”
3
Website widget resolves questions
ai_action
“a custom website widget - powered by ElevenLabs resolves frequent questions through chat or voice, reducing residents' need to call the city directly”
4
Subagents route to right LLM
routing
“visual workflows and Subagents assign the right LLM to each task, reducing latency and compute spend while keeping knowledge access targeted”
Reported outcome

Midland projects 7,000 fewer missed calls per month as overflow routes to Jacky instead of voicemail.
A site-wide widget is expected to deflect routine questions before they reach the phone queue, rapid language expansion improves community inclusion, and a privacy-first deployment maintains citizen trust.

Reported metrics
Missed calls per month (projected reduction)7,000 fewer missed calls per month
Inbound calls per day (baseline)more than 3,000
Latency and compute spendreducing latency and compute spend
Routine question deflectionexpected to deflect routine questions
Show all 5 reported metrics
missed calls per month (projected reduction)7,000 fewer missed calls per month
inbound calls per day (baseline)more than 3,000
latency and compute spendreducing latency and compute spend
routine question deflectionexpected to deflect routine questions
response quality / latencyfewer missed calls, faster responses, and broader language access
Reported stack
ElevenLabs AgentsSubagents
Source
https://elevenlabs.io/blog/city-of-midland
Read source ↗

Frequently asked questions

What did this team achieve with this AI workflow?

Midland projects 7,000 fewer missed calls per month as overflow routes to Jacky instead of voicemail.

What tools did this team use?

ElevenLabs Agents, Subagents.

What results were reported?

Missed calls per month (projected reduction): 7,000 fewer missed calls per month; Inbound calls per day (baseline): more than 3,000; Latency and compute spend: reducing latency and compute spend; Routine question deflection: expected to deflect routine questions (source-reported, not independently verified).

How is this customer support AI workflow structured?

Overflow call routed to Jacky → Jacky handles call with voice AI → Website widget resolves questions → Subagents route to right LLM.

Related customer support cases