The State of AI Agents in City Services: Insights & Strategies for Local Government Leaders

Botco AI Webinar Series

Speakers

Anu Shukla – Co-founder, Botco AI
Kari Johnson – Chief Data & Analytics Officer, City of Scottsdale, AZ
Rebecca Clyde – Co-founder & CEO, Botco AI

Opening

Anu Shukla:
Good morning and good afternoon, everyone. Welcome to Botco AI’s webinar series: The State of AI Agents in City Services. It’s a timely topic—government has been in the news a lot lately, but day-to-day city services continue to run.

We planned to have Wendy as a speaker, but she had an emergency. We’re very fortunate to be joined by Kari Johnson, Chief Data & Analytics Officer for the City of Scottsdale, Arizona. Kari is a data and AI leader, a champion of resident-centric government, and brings 15+ years of local government experience. She leads Scottsdale’s data and AI strategy, governance, and innovation to improve resident services, efficiency, transparency, and accountability. Welcome, Kari.

We’re also joined by Rebecca Clyde, co-founder and CEO of Botco AI, and long-time tech leader. Earlier in her career at Intel she focused on GTM strategy, and she later founded the agency Ideas Collide in Scottsdale before co-founding Botco AI.

I’m Anu Shukla, co-founder at Botco AI and serial tech entrepreneur. I joined Rebecca on this AI journey five to six years ago to apply AI to customer engagement and enterprise CRM.

About Botco AI

Anu Shukla:
Botco AI is a no-code SaaS platform to build, train, and deploy secure chatbots and agents. We integrate with CRM, CMS, EMR and more; we’re SOC 2 and HIPAA compliant. Our fine-tuned models are designed to minimize hallucinations—critical for regulated, mission-driven use cases. We typically reach ~98% answer accuracy within weeks of implementation.

Today we’re releasing our report, The State of AI Agents in City Services, with strategies for local government leaders based on what cities across the U.S. are doing. We surveyed 100 U.S. municipal leaders in August 2025 via Pollfish to understand adoption, ROI, and plans.

Audience Poll (Live)

Anu Shukla:
While we discuss, we ran a quick poll on your stage of AI adoption. Results align closely with our survey: roughly 27% fully implemented, ~33% completed a pilot (about 50% total adoption), ~27% exploring, and a small group with no current plans.

Rebecca Clyde:
We see very mature adopters—Scottsdale is a great example—and others earlier in the journey. It’s helpful to understand where you are to plan next steps.

Scottsdale’s Approach

Kari Johnson:
In Scottsdale, we adopted AI—and AI governance—early. We have several initiatives live and in motion. Internally, we launched a bot named Sadie (after a beloved former mayor’s dog). Sadie helps staff open work orders, draft emails, find buildings and maps, and surface policies and documentation from SharePoint—very broad knowledge.

We’re continually expanding Sadie’s skills. For example, we’re exploring reminders (“Remind me in two days to change my password”). We also plan a resident-facing chatbot that can hold conversational interactions: “I’m visiting with a five-year-old—what can we do this weekend?” It will pull from calendars to suggest relevant events, and integrate with other city services so residents don’t have to click through multiple sites.

Meeting Consumer-Grade Expectations

Anu Shukla:
Residents expect the 24/7, seamless experience they get from consumer apps. Budgets and staffing make this hard. How can cities move in that direction?

Kari Johnson:
Government is held to different standards and can’t move at “warp speed.” Governance is essential—before rolling out tools, even internally. Create a cross-functional group (IT, HR, Legal, Risk, etc.) to manage and mitigate risk, ensure ethical use, and build trust guardrails. We’re mission-based, not revenue-driven, so we have to approach AI from that lens. Survey residents—ask how AI could help them.

Rebecca Clyde:
Include residents in UX testing. Involve people with lived experience so the solution is “of the community.” In Arizona, community input led to adding heat-shelter and water-station lookups: in under 20 seconds, a resident can find the nearest relief location during extreme heat.

Adoption Landscape & Why Act Now

Anu Shukla:
Our survey shows 50%+ have implemented or are considering chatbots; a meaningful share still has no plans. Advice to the latter?

Rebecca Clyde:
Talk to leaders like Kari. Cities that deliver consumer-grade service become more livable and business-friendly. Scottsdale’s innovation attracts residents and businesses.

Kari Johnson:
AI is a transformation on par with the late-90s internet. It’s not a fad. Residents expect services they hear about elsewhere, and your staff must learn to work differently. Not bringing AI into your culture disadvantages your workforce. Like teaching computer skills in the 90s, this is foundational.

Biggest Engagement Challenges & AI’s Role

Anu Shukla:
Top challenges reported: high call-center volume (12%), slow response times (24%), lack of real-time updates, budget constraints (26%), and multilingual support (11%). How can AI help?

Rebecca Clyde:
Multilingual support is a major value—many communities are linguistically diverse. AI can provide equitable access across services (water, trash, parks & rec) with strong multilingual fluency.

Kari Johnson:
Internally, staff also need translation—for voicemails or emails in other languages—so they can respond accurately and quickly. Multilingual support is essential everywhere.

Privacy & Security

Anu Shukla:
How does Botco protect citizen data while offering personalization?

Rebecca Clyde:
We encrypt sensitive fields (e.g., addresses) with dedicated keys, enforce role-based access controls, and can delete PHI/PII immediately based on use case. Controls can be tuned per scenario to safeguard personal data.

Features Cities Value Most

Anu Shukla:
Most-valued chatbot capabilities: 24/7 availability, real-time updates on services/closures, integration with city systems, multilingual support, and automated appointment scheduling. Emergency notifications and disaster response are critical.

A mid-sized Texas city received 300–400 weekly calls about trash: pickup days, recycling rules, missed pickup. A chatbot deflected many of those calls successfully.

Kari Johnson:
Utility over novelty is right. 311 and call-center logs are a treasure trove for prioritization. Cities are like 35 distinct “businesses” that must sometimes share data. Agents can be trained to answer specific common questions (“When is my bulk trash day?”) without forcing residents to decipher PDFs or color-coded maps. Timely, correct information matters.

Anu Shukla:
City information changes frequently—officials, offices, policies, and rates. Agents should continuously retrain against reliable sources to stay current.

Rebecca Clyde:
Exactly—real-time updates go hand-in-hand with emergency response. In disaster-prone communities, continuous retraining and audit trails dramatically improved outcomes during tornadoes/hurricanes.

Bridging the Expectation vs. Capability Gap

Kari Johnson:
AI helps reduce tech debt. Rather than heavy data migrations, focus on clean, accessible data that agents can reference. Follow strong data hygiene and publish confidently.

Rebecca Clyde:
Mechanically, platforms like ours crawl designated sources (hourly, nightly, etc.) and retrain agents accordingly. We show answers and sources with timestamps for auditability and trust, so users don’t have to “treasure hunt” through nested calendars.

Reducing Hallucinations

Anu Shukla:
How does Botco reduce hallucinations?

Rebecca Clyde:
We use an AI Playground to test source coverage and answer fidelity. Multi-layer agents fact-check responses, and we provide audit trails (source, time) to ensure accountability. We offer the playground free to communities because it’s core to ethical AI.

Staffing & Change Management

Audience Q: How should cities prepare for staffing adjustments as AI is adopted?

Kari Johnson:
In Scottsdale, our administrative regulation states AI is not intended to replace humans. Leadership consistently frames AI as a force multiplier that performs certain tasks faster—but it doesn’t “know more” than we do, nor understand context like humans. Culture matters: publish policy, socialize it, and model usage at the top.

We launched Sadie with intentional change management: naming, friendly messaging (“Sadie can fetch this,” “learn new tricks”), and short training. We embraced Microsoft Copilot’s arrival by providing simple guidance and security training. We built an AI Resource Center in SharePoint with approved tools, use cases, licensing contacts, and beginner-friendly training. Executive sponsorship is essential; we continue to meet people where they are.

Barriers & How to Overcome Them

Anu Shukla:
Common barriers: initial cost, accuracy/reliability concerns, and staff anxieties. On cost: can you afford not to try? Once you see ROI, expansion becomes obvious.

To lower the barrier, Botco offers a free, ready-to-train “Trash Bot” focused on waste and recycling—an area with immediate call deflection and resident value. It’s quick to launch using your data.

Rebecca Clyde:
Rules (e.g., “Do you take glass?”) vary and change. A bot helps residents and visitors get the right answer instantly. We’ve seen call-center deflection and employee productivity improvements that more than cover costs once adoption takes hold.

Multi-Agent Orchestration & Costs

Audience Q: We need multiple agents for different tasks and want them to coordinate. How do others handle orchestration and cost?

Rebecca Clyde:
Start with small, narrowly trained agents to reduce error risk, then compose them into a larger “mega-agent.” Orchestration requires clear APIs and security/communications protocols among agents. This is where platform design matters.

Kari Johnson:
Choose a platform that can host all your agents (like Botco), with connectors to Microsoft, Google, ServiceNow, etc. That centralization simplifies governance and cost control (predictable SaaS) versus ad-hoc LLM usage that can spike costs. Prefer low/no-code where possible.

Closing & Key Takeaways

Anu Shukla:
We’re out of time. Everyone who registered will receive the report. Don’t risk falling behind—pilot thoughtfully and bring your people along. Great starting point: the free Trash Bot to demonstrate quick value. We’re also partnering on resident navigation, licensing, fees, renewals—integrated across systems for a seamless experience.

Thank you, Kari and Rebecca, and thanks to all who joined.

Kari Johnson:
Thank you for having me.

Rebecca Clyde:
Thanks, everyone. Take care.