← Cases
United StatesPublic sector2024 – ongoing

Montgomery County, Maryland

Monty 2.0 resident chatbot

Montgomery County: Monty 2.0 covers 3,000 topics and lowers unanswered-query rates to 10–15%

WorkingVendor's case studyVendor ↗

Problem

The county's first chatbot covered about 20 topics. Officials also worried that generative AI could present false information as fact.

Approach

The county built Monty 2.0 with Microsoft and Zammo.ai on Azure OpenAI Service and Azure AI Search. It answers only from the county's knowledge base, translates replies and uses mapping data for address-specific answers. A seven-month beta with a resident focus group started in March 2024.

Results

35–45% → 10–15%Unanswered-query rate, before and afterVendor's case study[1]
3,000Topics covered at full launch, versus 20 for the original MontyVendor's case study[1]
140Languages supported through automatic translationVendor's case study[1]

Takeaway

Unanswered queries fell from a 35-45% range to 10-15%, as the county reports. The 20,000 conversations counted are from the beta period.

Categories

Business function
Customer service
Evidence
Vendor's case study

Sources

  1. Montgomery County customer story
    Microsoft26 Mar 2025Vendor publication

Source summaries

[1]Microsoft26 Mar 2025

The county rebuilt its resident chatbot on Azure OpenAI with answers limited to its own knowledge base. Microsoft reported topic coverage and the drop in unanswered queries.

Full details are in the original sources.

Added 25 Oct 2026 · Updated 25 Oct 2026
Open case page ↗