From Chat to Voice Why Phone Lines are the Next Frontier

If web chat was the first wave of AI powered resident interaction, voice is the next and perhaps the most disruptive. For many agencies, the phone remains the primary front door because large portions of the population still default to a call when something goes wrong or when trust in digital channels is fragile. Yet the mechanics of that experience have barely changed "Press 1 for this, press 2 for that" even as call volumes swell and budgets tighten.​​

Citibot's AI Voice is designed to replace or augment those traditional IVR trees with a conversational layer that sits on top of the existing Citibot platform. Residents call the same numbers they already know for 311 or departmental lines and instead of navigating a rigid menu, they simply say what they need in their own words and are either answered directly, routed to the right queue, or invited to submit a service request. Behind the scenes, Citibot maps those natural language queries to the same knowledge base and workflows that power web chat and text, so agencies are not juggling separate sets of rules.​

Bratton defines AI Voice as a “high-quality, AI-powered agent” designed for more than just novelty: the goal is total parity with human operators on every routine call.

Citibot can even point existing phone numbers to the AI platform without changing infrastructure, creating a soft transition where residents experience better service without needing to learn a new entry point.​

Consider the pressure on a public works department during a missed pickup event or a storm. With AI Voice running on a main number, hundreds of residents can report issues, receive confirmation numbers, and in some cases be offered immediate scheduling updates, all without waiting on hold.

Data, Sentiment, and the New Analytics Fabric

For years, call centers and 311 systems produced reports sorted by category and count, but they rarely captured the emotional texture of what residents were saying in real time. Citibot's upgraded analytics dashboard is an attempt to close that gap by overlaying sentiment, topics, and time frames on top of the raw conversation stream.​​

Jordan Schinstock describes the ambition this way: “By capturing everything from compliments to complaints and the middle majority, agencies gain a 360-degree view of resident sentiment.”

The objective is to move agencies beyond anecdotes, one angry email, one viral social media thread and toward decisions rooted in thousands of AI mediated conversations happening every week.​​

The new dashboard, provided at no additional cost to Citibot customers, turns that stream into operational insight. Leaders can see when spikes in trash questions map to missed routes, when confusion around a new ordinance is concentrated in a particular neighborhood, or when website content is failing to answer frequent queries, often within days of a change. Over time, those patterns become an early warning system for friction that might otherwise appear months later in satisfaction surveys or council meetings.​

Denver's experience offers a glimpse of what this looks like in practice. 

Laura Dunwoody, the city's Director of Technology Services for Resident Experience, explains that with Citibot, "We can now give our agency customers specific questions residents are asking and pinpoint exactly what website content needs updating, making our discussions shorter and more productive." 

Instead of broad debates about whether a department is doing enough outreach, her team brings transcripts and trends that show exactly where answers are failing and how updated language changes the flow.​​

This feedback loop changes internal culture. Meetings that once relied on hunches become grounded in the language residents actually use and the complaints they repeat. IT and communications staff gain leverage because they can demonstrate that a fix to one paragraph on a page or an adjustment to a workflow will likely reduce a measurable slice of calls or chats. In effect, Citibot's analytics recast AI not only as a front line service but as a kind of observatory for the lived experience of public services.​

Refresh and the Battle Against "Old as Dirt" Content

Every digital leader in government knows the problem of content drift. Pages created years apart contradict each other. PDFs linger with expired dates. Different departments speak in different voices about the same program. Residents bump into broken links and outdated instructions and draw the only conclusion they can: this agency is not paying attention.​​

Refresh uses AI to scan government websites and PDFs for outdated content, inaccuracies, conflicting information, broken links, and accessibility gaps, all while scoring pages across four pillars: accuracy, current, persona, and accessibility. Rather than relying on staff to manually audit hundreds or thousands of pages, agencies receive prioritized lists of issues and specific locations to fix.​​

The timing is not incidental. New accessibility regulations will require cities and counties above certain population thresholds to show progress toward WCAG 2.1 compliance by 2027, with large jurisdictions needing to demonstrate they are "well on their way" by April of this year. 

Practical Roadmap for 2026 and Beyond

• First, establish a strong AI chat foundation across web and text focusing on one or two high volume service areas such as public works or permitting and using Citibot's full service approach to minimize IT lift. Early months should emphasize content curation and governance ensuring that the knowledge base reflects current policies and that internal teams understand how residents are using the new channel.​

• Second, take a deliberate deep dive into the enhanced analytics dashboard and start weaving sentiment and topic trends into regular leadership conversations. Use those insights to review content at scale and prioritize website edits, policy clarifications, and internal training rather than waiting for annual reports or survey cycles. As Schinstock’s framing suggests, the goal is to move from guessing to knowing, and from isolated anecdotes to patterns that can guide investment

• Third, expand into AI Voice by routing existing high volume phone lines into the Citibot platform, starting with after hours coverage or specific departments that are struggling with call volume. Communicate clearly with residents about the change, emphasizing that they can still reach staff when needed, but that the agency is introducing a faster option for routine questions and requests.​

• Finally, bring chat, Voice, analytics, and Refresh together as a single resident experience layer rather than separate tools. When your AI assistant, phone lines, and website all run on the same Citibot platform, you can use WCAG 2.1 oriented scans and persona scoring to keep content accurate and accessible while every channel delivers consistent answers and eliminates the old as dirt contradictions that frustrate residents and confuse AI alike.

Throughout this progression, the central discipline is to treat AI not as a black box, but as a strategic extension of institutional record and trust. This requires owning the data, curating the content, and measuring success by outcomes that matter - faster answers, fewer handoffs, and clearer explanations - rather than technical novelty alone.

Citibot 's 2026 platform provides the essential fabric for scalability and resilience. By automating complex interactions and ensuring every response is accurate and accountable, the platform removes the operational bottlenecks that stifle expansion. Cities and government agencies that move early to integrate this technology will not only meet rising expectations but will unlock the efficiency and responsiveness required to redefine their market position and accelerate long-term growth.