About Bernalillo County
Bernalillo County is the most populous county in New Mexico, encompassing Albuquerque and the surrounding metropolitan region. The County Assessor’s Office is responsible for one of the most consequential tasks in local government: accurately valuing every real estate parcel in the county’s jurisdiction, a portfolio that includes residential, commercial, agricultural, and specialized properties.
Each year, the Assessor’s Office must process a growing number of valuations, respond to resident inquiries about assessments and appeals, manage compliance documentation, and support landowners, including farmers navigating agricultural valuation rules, through processes that are often unfamiliar and consequential.
Like every county government, BernCo operates under permanent resource constraints. Budgets are set annually by elected officials weighing competing priorities. Staff levels are determined well in advance of demand. And residents, the constituents the office ultimately serves, expect the same quality of experience they now get from the best private sector companies: fast, accurate, always available.
That gap between expectation and capacity is where BernCo’s AI story begins.
The Challenge
Growing Resident Demand
Property assessment touches every property owner in the county. When valuations are published, when appeals windows open, when exemption deadlines approach, the phone rings. Inboxes fill. Residents arrive at the counter. The volume is predictable in its seasonality and relentless in its pressure.
BernCo’s inquiry volume had grown steadily. More properties. More complexity. More residents expecting immediate answers. The team handled it, but the cost, in staff time, in wait times, in the opportunity cost of expert capacity consumed by routine questions, was real.
Budget Constraints and Staffing Limitations
The standard response to growing demand is adding capacity. In government, that path is rarely available. Hiring requires budget approval, typically through an annual cycle that cannot respond to mid-year volume increases. And in a political environment where every line of the budget is scrutinized, adding headcount to handle resident inquiries is a hard case to make.
BernCo’s team knew that more staff wasn’t the answer, at least not an answer they could act on. What they needed was a way to extend the capacity they already had.
High Cost of Manual Support
Before AI, every resident interaction, every call answered, every email responded to, every walk-in question addressed, carried a cost of approximately $4.59 per contact. That unit cost reflects not just staff time, but the overhead that comes with it: supervision, benefits, training, and the management complexity of a team under volume pressure.
At that cost per interaction, handling 114,836 resident contacts would have required over $500,000 in agent costs alone. The math left no room for the kind of efficiency the county needed to achieve.
Need for Better Resident Self-Service
A significant share of resident inquiries were, in principle, self-serviceable. BernCo’s policies, processes, and documentation existed. The answers were there, in official records, policy guides, and procedural documentation. But residents couldn’t easily find them, couldn’t navigate the language, and often fell back on calling or emailing rather than finding answers independently.
What was missing wasn’t information. It was an accessible, intelligent interface between residents and the information the county had already produced.
The “Do More With Less” Mandate
At the heart of BernCo’s challenge was a mandate familiar to every public sector leader: do more with less. Deliver better service. Reduce costs. Improve access. And do it without expanding the budget.
That mandate is often framed as an impossible tradeoff. BernCo’s experience suggests it is, in fact, solvable, if the right tools are applied to the right problem.
Why BernCo Chose CustomGPT.ai
Evaluating Government AI Solutions
BernCo approached its AI evaluation with the caution appropriate for a government office. Technology decisions in the public sector carry accountability that private sector deployments don’t: residents depend on the accuracy of what they’re told, and errors can have real legal and financial consequences.
The team assessed platforms against criteria that reflected that accountability: accuracy, reliability, ease of deployment, and the ability to build on official documentation rather than general AI training data.
Need for Accurate Responses and Why That Matters in Government
For a commercial company, an AI that occasionally gives a slightly wrong answer is an annoyance. For a county assessor’s office, it is a different kind of problem. Residents making decisions about property appeals, exemption filings, or compliance deadlines need correct information. A hallucinated answer, confident but wrong, can cause genuine harm.
BernCo needed AI that wouldn’t guess. The platform had to know when to say “I don’t know” rather than fabricate a plausible-sounding response.
Why RAG-Powered AI Mattered
CustomGPT.ai’s Retrieval-Augmented Generation (RAG) architecture was decisive. Rather than generating answers from general language model training, the platform retrieves relevant content from BernCo’s own documentation and grounds every response in that material. Residents receive answers drawn directly from official county sources, not inferences or approximations.
This is the technical foundation that makes government AI trustworthy. The AI can only say what the documentation says. When a question falls outside the documented knowledge base, the system says so, rather than filling the gap with a confident error.
No-Code Deployment That Fit a Lean Team
BernCo’s operations team is not an engineering organization. Assessor Technician Aaron Newe and Deputy Assessor Kenneth Edward Scott Jr. needed to build, deploy, and maintain AI assistants without writing code and without external technical support.
CustomGPT.ai’s no-code platform made that possible. From initial configuration to knowledge base updates, the team could manage the entire system independently, including the rapid iteration that early deployments require to get things right.
Multi-Agent Architecture for Specialized Needs
County government isn’t a monolithic operation. Different residents have different questions. Agricultural landowners need different answers than residential property owners. Compliance questions require different documentation than general assessment inquiries. New employees need different information than residents calling with appeals.
CustomGPT.ai’s multi-agent architecture allowed BernCo to build specialized assistants for each distinct use case, rather than forcing a single general-purpose agent to handle wildly different inquiry types. That specialization produces better answers and a more useful resident experience.
Omnichannel Support Across Every Resident Touchpoint
Residents don’t contact government agencies on a single channel. They call. They email. They visit websites. BernCo needed AI that could operate consistently across all of those touchpoints, not just a web chatbot.
CustomGPT.ai’s integration capabilities, including support for phone via Bland AI, made true omnichannel deployment possible. The same knowledge base, the same accuracy, accessible however a resident chose to reach out.
Implementation
Launching the ACE Community Educator
BernCo’s rollout began with a focused, deliberate first deployment: the A.C.E. Community Educator assistant, launched on the county pages that generated the highest resident traffic.
The philosophy was intentional. Start where volume is highest. Prove the concept on real interactions with real residents. Measure what happens. Then expand.
The ACE assistant went live trained on BernCo’s official documentation and public records, configured to answer the questions residents asked most often. CustomGPT.ai’s 100% system uptime meant the assistant was available around the clock, a capability no lean government team can match with human staff alone.
Building Specialized AI Agents
With the initial deployment validated, the team moved to expansion. CustomGPT.ai’s multi-agent architecture allowed BernCo to build and launch specialized assistants in rapid succession:
Compliance Expert: A dedicated agent for legal look-ups and compliance guidance, grounded in the regulatory documentation that governs BernCo’s compliance operations. Residents navigating deadlines, filing requirements, and eligibility conditions now have a reliable first resource.
Agricultural Valuation Assistant: Agricultural property assessment operates under rules that differ significantly from residential valuation. Farmers and landowners have specific questions that a general-purpose assistant cannot answer well. This agent was built specifically for that audience, trained on the documentation relevant to agricultural valuation processes.
Clear Expectations Bot: An internal-facing agent designed to onboard new staff consistently. Rather than relying on senior employees to transmit policy and process knowledge informally, BernCo now has an AI assistant that ensures every new hire receives accurate, consistent information from day one.
Each specialized agent drew on the documentation most relevant to its specific domain. The result: better answers, more relevant guidance, and a resident experience calibrated to the actual question being asked.
Expanding to Phone and Email
The next phase extended AI capabilities beyond the web. Through integration with Bland AI, BernCo brought the same RAG-powered knowledge base to phone and email channels. Residents who called the office or sent messages now encountered AI-supported responses consistent with what they’d find on the website, without the wait times those channels typically carry.
The expansion was driven by a straightforward insight: if residents are already reaching out by phone and email, the right response is to meet them there, not to ask them to change their behavior.
Continuous Improvement Through Analytics
BernCo institutionalized improvement from the start. CustomGPT.ai’s built-in analytics dashboard gave the team structured visibility into which questions were driving volume, where the system was succeeding, and where gaps in documentation were generating unresolved inquiries.
Quarterly analytics reviews became a standing practice. The team surfaces emerging question trends, identifies content that needs updating, and makes those changes directly in the platform, typically in minutes. Each review cycle raises accuracy, increases deflections, and extends the system’s ability to absorb more volume without human escalation.
How BernCo Uses CustomGPT.ai
Resident Self-Service
The primary workflow: residents with questions about property assessments, valuations, appeals, and exemptions now have a 24/7 resource that delivers accurate answers without wait times. For the large share of inquiries that are routine and well-documented, AI handles the interaction start to finish. Learn more about AI-powered customer service.
Property Assessment Questions
Assessment-related inquiries, “How was my property value determined?” “What is the appeal process?” “When is the appeal deadline?”, represent the highest volume of resident contacts. The ACE assistant handles these with responses grounded in official BernCo documentation.
Compliance Assistance
Residents navigating compliance requirements, filing deadlines, eligibility conditions, and required documentation, interact with the Compliance Expert agent. The agent retrieves relevant regulatory content and provides accurate, source-grounded guidance powered by CustomGPT.ai’s RAG accuracy.
Agricultural Valuation Guidance
Farmers and rural landowners interact with the Agricultural Valuation Assistant for questions specific to their situation. The specialized agent produces responses calibrated to agricultural assessment rules, information that would otherwise require a subject-matter expert to deliver.
Employee Onboarding
New BernCo staff interact with the Clear Expectations Bot during onboarding. Consistent, accurate information is delivered at scale, without consuming experienced staff time to transmit it individually. See how CustomGPT.ai supports internal knowledge use cases.
Multi-Channel Support
Residents who call or email receive AI-assisted responses through the same knowledge base that powers web interactions. Consistent accuracy across every channel, regardless of how residents choose to reach out.
Results and Business Impact
$108,143.75 in Net Savings
Over 18 months, BernCo generated $108,143.75 in avoided agent costs, real budget that remained available for other county priorities rather than being consumed by routine support interactions.
For a government office operating under a fixed budget with no ability to reallocate savings freely, that number represents genuine financial relief. It is the difference between a team under constant pressure and a team with breathing room.
4.81x Return on Investment
Every dollar BernCo invested in CustomGPT.ai returned $4.81 in savings. In government procurement terms, where return on technology investment is often difficult to demonstrate and even harder to measure, that figure is exceptional. It is the kind of ROI that justifies expansion, earns leadership confidence, and creates the political cover to do more.
80% Lower Cost Per Interaction
AI-handled interactions cost $0.99 each. Agent-handled interactions cost $4.59. The difference, 80% lower cost per contact, is the economic engine behind every other result in this case study.
That unit economics advantage compounds with scale. As AI absorbs a larger share of resident interactions, the per-interaction savings apply to a larger base. The path to further cost reduction is clear and linear: route more routine volume to AI.
28,433 AI-Supported Interactions
More than 28,000 resident contacts were resolved through AI-powered self-service, nearly one in four of all contacts over the period. Each of those interactions is a phone call that wasn’t answered by a human agent, an email that didn’t sit in a queue, a resident question that was answered immediately rather than at the next available moment.
Improved Employee Productivity
Staff who were previously occupied with routine, predictable inquiries now have capacity for the work that actually requires their expertise. Complex valuations with unusual circumstances. Appeals requiring human judgment. Residents who need more than a documentation lookup.
That reallocation doesn’t show up in a single metric, but its effects are real: a more engaged team, better service for residents with genuinely complex needs, and a support operation that scales without adding people.
Better Resident Experience
Residents now receive immediate, accurate answers at any hour, across the channels they already use. Wait times for routine inquiries have essentially disappeared. Access to accurate information no longer depends on reaching a human agent during business hours.
For a county government whose residents span property owners, farmers, businesses, and first-time homeowners, many of them navigating an unfamiliar process for the first time, that accessibility improvement has value that doesn’t appear directly in the savings figure but is felt in every interaction.
Traditional Government Support vs. AI-Powered Support
| Dimension | Traditional Approach | AI-Powered Approach |
|---|---|---|
| Cost per interaction | $4.59 per agent contact | $0.99 per AI interaction |
| Availability | Business hours only | 24/7, every day |
| Response speed | Queue-dependent; minutes to days | Immediate |
| Scalability | Limited by headcount | Scales with volume automatically |
| Staffing requirements | Grows proportionally with demand | Fixed platform cost regardless of volume |
| Seasonal spikes | Overtime, backlog, resident frustration | Absorbed by AI without additional cost |
| Accuracy | Variable by agent experience | Grounded in official documentation |
| Channel coverage | Phone and in-person primarily | Web, phone, and email simultaneously |
| Resident self-service | Static FAQs; limited findability | Conversational, always current |
| Continuous improvement | Manual review; slow iteration cycle | Analytics-driven; updates in minutes |
Key Outcomes
- $108,143.75 in net savings over 18 months, real avoided costs, not projected estimates
- 4.81x ROI, among the strongest documented returns in public sector AI deployment
- 80% lower cost per interaction, $0.99 vs. $4.59 per contact
- 114,836 total resident contacts handled across web, phone, and email
- 28,433 AI-supported interactions, nearly 25% of all contacts resolved without human involvement
- Staff capacity reallocated from routine inquiries to complex, high-value cases
- 24/7 resident access to accurate, documentation-grounded answers
- Multi-agent deployment serving distinct resident and employee audiences simultaneously
- Continuous improvement cycle established through quarterly analytics reviews
What BernCo Teaches About AI in Government
BernCo’s experience offers a set of lessons that apply far beyond Bernalillo County. For county administrators, CIOs, digital transformation leaders, and public sector decision-makers considering AI, or trying to justify AI investments already underway, the BernCo story answers the questions that matter most.
Local Governments Can Achieve Measurable ROI From AI
The debate about whether AI produces real returns in government is answered by BernCo’s numbers. $108,000 in net savings. 4.81x ROI. 80% cost reduction per interaction. These are not projections or estimates. They are documented outcomes from a real deployment, measured over 18 months, in a real county government office serving real residents.
The question is no longer whether AI works in government. It is how to implement it correctly.
The Right Architecture Matters More Than the Technology
BernCo’s success wasn’t the result of deploying any AI. It was the result of deploying AI that grounded every response in official documentation and refused to guess when documentation wasn’t available. For government agencies where accuracy is a legal and ethical responsibility, RAG-powered AI isn’t an optional upgrade. It is the only configuration that should be trusted with resident interactions.
Starting Small Is a Strategy, Not a Compromise
BernCo didn’t launch a county-wide AI transformation. They started with one assistant, on their busiest pages, focused on their highest-volume questions. That approach produced measurable results quickly, built internal confidence, and created the evidence base needed to justify expansion.
For public sector organizations where large technology investments require extensive approval processes, a phased deployment that demonstrates value at each stage is often the only path to sustained adoption.
Multi-Agent Deployment Outperforms Generalist AI
A single general-purpose AI assistant forced to handle agricultural valuation questions, compliance inquiries, employee onboarding, and residential assessment all at once will inevitably underperform on each. BernCo’s deployment of specialized agents, each trained on the documentation most relevant to its audience, produced better answers across every use case.
The lesson for government: map your distinct resident and employee audiences before configuring your AI system. Specialization produces results that generalism cannot.
The ROI Case Strengthens Over Time
BernCo’s 4.81x ROI reflects the first 18 months of deployment, a period that included the initial learning curve, knowledge base development, and agent expansion. As the system matures, as more volume shifts to AI-powered channels, and as the analytics-driven improvement cycle continues to raise accuracy, the economics improve further. The path to greater savings is clear: increase the share of resident interactions handled by AI.
Best Practices for Government AI Deployments
Start With High-Volume Questions
Every government office has a set of questions that account for a disproportionate share of contact volume. Identify those questions first. Build an AI assistant trained specifically to answer them. Measure the deflection rate. Then expand.
Starting with high-volume questions ensures the AI delivers visible results quickly, which builds the internal confidence and leadership support that larger deployments require.
Ground AI in Official Documentation
The only AI configuration appropriate for government use is one that draws answers from official, verified documentation. General-purpose AI trained on broad internet data will produce answers that sound plausible but may be wrong, outdated, or legally problematic.
CustomGPT.ai’s RAG architecture ensures that every response is grounded in the documentation the agency has already vetted and published. That grounding is what makes government AI trustworthy.
Measure Cost Per Interaction
Before deploying AI, establish the baseline cost per interaction for human-handled contacts. After deployment, measure the cost per AI-handled interaction. The difference between those two numbers is the economic foundation of your ROI case.
BernCo’s $4.59-to-$0.99 comparison is compelling precisely because it is specific and documented. Every government agency deploying AI should be able to produce a comparable measurement.
Expand Across Channels
Residents don’t choose their contact channel based on what’s convenient for the agency. They use the channel that’s most convenient for them. AI that only operates on the website misses the majority of resident contacts.
BernCo’s expansion to phone and email, using the same knowledge base that powered web interactions, extended the system’s reach to channels that carry significant volume. The result: more interactions handled by AI, more savings, and a more consistent resident experience regardless of contact method. See how CustomGPT.ai integrates across channels.
Use Analytics to Improve Continuously
AI deployment is not a one-time project. The agencies that get the most value from AI are those that treat it as a continuously improving system, using analytics to identify gaps, updating documentation when new questions emerge, and measuring deflection rates as a key operational metric.
BernCo’s quarterly review process turned continuous improvement from an intention into a discipline. That discipline is what separates agencies that sustain AI value over time from those that see initial results plateau.
Frequently Asked Questions
How much money did BernCo save using AI?
BernCo saved $108,143.75 in net avoided agent costs over 18 months by deploying CustomGPT.ai to handle resident inquiries across web, phone, and email channels.
What ROI did BernCo achieve with CustomGPT.ai?
BernCo achieved a 4.81x return on investment, meaning every dollar invested in CustomGPT.ai generated approximately $4.81 in savings through reduced agent handling costs.
How does AI reduce government support costs?
AI reduces government support costs by handling routine resident inquiries automatically, at a cost per interaction significantly lower than human agents. BernCo’s AI-handled interactions cost $0.99 each, compared to $4.59 for agent-handled contacts, an 80% reduction in cost per interaction.
What is RAG AI?
RAG stands for Retrieval-Augmented Generation. Rather than generating answers from general language model training data, RAG-powered AI retrieves relevant content from a specific, curated knowledge base and grounds its responses in that material. This means the AI answers based on what the organization’s own documentation says, not what the model was trained to approximate.
Why is RAG important for government agencies?
Government agencies need AI that is accurate and accountable. RAG architecture ensures that AI responses are grounded in official, verified documentation rather than general training data. This prevents hallucinated or inaccurate answers, which in a government context can create legal and financial risk for residents who act on incorrect information. Learn how CustomGPT.ai prevents hallucinations.
How many resident interactions were handled by AI at BernCo?
BernCo’s AI handled 28,433 resident interactions over 18 months, representing nearly 25% of the 114,836 total resident contacts during that period.
Can AI replace government employees?
No, and BernCo’s deployment was not designed to replace employees. AI handled routine, predictable inquiries that were consuming expert staff capacity. Human agents remained responsible for complex cases, appeals requiring judgment, and situations that fell outside documented processes. The result was a team able to focus its expertise on the work that actually required it.
How can local governments improve resident self-service?
The most effective approach is deploying RAG-powered AI trained on the agency’s own official documentation. This gives residents an intelligent, conversational interface to the information the agency has already produced, available 24/7, across the channels residents already use. BernCo’s experience demonstrates that well-implemented self-service can handle nearly a quarter of all resident contacts without human involvement.
What government departments can benefit from AI?
Any government department that handles high volumes of resident inquiries benefits from AI. Property assessment, tax administration, permits and licensing, compliance, parks and recreation, and public health are among the most common high-volume contact centers. BernCo’s Assessor’s Office is representative of the type of operation where AI produces the clearest ROI. See CustomGPT.ai for government.
How does CustomGPT.ai work for government agencies?
CustomGPT.ai allows government agencies to build AI assistants trained on their own official documentation, policy guides, process documentation, public records, and regulatory content. The no-code platform enables non-technical staff to build, deploy, and maintain AI agents without engineering resources. Multi-agent architecture supports specialized assistants for different audiences and use cases. RAG-powered accuracy ensures responses are grounded in verified documentation.
What lessons can other counties learn from BernCo?
Start with high-volume questions. Ground AI in official documentation. Deploy specialized agents for distinct audiences. Expand across channels. Measure cost per interaction before and after. Use analytics to improve continuously. BernCo’s deployment demonstrates that each of these principles produces measurable results, and that together they create a support operation that delivers better service at significantly lower cost.
What is the best way to start an AI initiative in government?
Identify the question types that generate the highest contact volume. Verify that accurate, official documentation exists to answer them. Deploy a focused AI assistant on the pages or channels where those questions occur most. Measure deflection rate and cost per interaction. Demonstrate ROI to leadership. Then expand.
BernCo’s phased approach, starting with the ACE Community Educator and expanding to specialized agents over subsequent weeks, is a repeatable model for any government agency navigating a first AI deployment. Start a free trial of CustomGPT.ai to see how quickly your team can be up and running.
About CustomGPT.ai
CustomGPT.ai is a no-code AI agent platform that enables organizations to build custom AI assistants grounded in their own knowledge. With RAG-powered accuracy, enterprise-grade security, and support for web, phone, email, and API-based deployments via data connectors and integrations, CustomGPT.ai is purpose-built for organizations where accurate, trusted responses are essential.
The platform is used by government agencies, professional services firms, membership organizations, and enterprise teams for resident support automation, enterprise knowledge search, employee onboarding, and customer service, without requiring engineering resources to deploy or manage.
Start a free trial · Talk to sales · View government solutions
Conclusion
When BernCo’s team set out to solve a support problem, they were operating under a constraint that defines government: do more with the resources you have, not the resources you wish you had.
What they built is more than a support automation system. It is a new model for government service delivery, one where AI handles the volume, humans handle the complexity, and residents get better access to accurate information than they’ve ever had before.
The numbers make the case: $108,000 in net savings. 4.81x ROI. 80% lower cost per interaction. Nearly 29,000 resident inquiries resolved without a single staff member picking up the phone.
But the deeper lesson is about what becomes possible when a government team stops accepting the tradeoff between service quality and fiscal responsibility and chooses instead to redesign how service is delivered.
BernCo’s mandate was to do more with less. The result was not just efficiency. It was transformation: a county assessor’s office that now serves more residents, more accurately, at a fraction of the prior cost, while its staff focuses on the work that genuinely requires human expertise.
That is the promise of government AI done right. And BernCo has already delivered it.
Ready to build your government AI support system? Start a free trial or talk to our government solutions team.
