The Complete Guide to Using AI in the Government Industry in Durham in 2025

Durham County's 2025 AI rollout shows why local governments across North Carolina should treat AI as operational infrastructure, not just an experiment: by deploying Moveworks' Copilot - branded locally as “DCoBot” - the County automated routine IT requests across 30+ departments serving 2,200+ employees, integrated the assistant into Microsoft Teams and ServiceNow, and within the first 30 days the bot handled a substantial number of routine inquiries, freeing help‑desk staff for higher‑value work and boosting employee productivity; learn the implementation details in the Durham County Moveworks case study and explore government‑grade features like Moveworks GovCloud on the Moveworks AI for Local Government page - practical staff training matters for adoption, so consider the Nucamp AI Essentials for Work syllabus to build prompt‑writing and operational skills for public sector teams.

"This transformation is not just about technology; it's about setting a new standard for public service in the digital age." - Chasity Locke, Assistant Director of IT Engagement and Communications

Durham beginners should know that the industry outlook for 2025 is mixed but actionable: empirical research finds AI adoption often correlates with firm growth, increased employment, and heightened innovation - especially in product development - so local government can harness AI to expand services and efficiency (Brookings study on AI effects on firms and workers); at the same time, generative models are already reshaping livelihoods - freelancers in more exposed occupations experienced a 2% decline in contract counts and a 5% drop in earnings - which signals real short‑term disruption for roles that mirror those freelance tasks (Brookings analysis of generative AI's labor impacts in the freelance market).

The practical takeaway for Durham: pair pilots with a clear reskilling plan and skills‑first hiring to capture AI's upside while protecting workers, and start by running focused pilots with measurable metrics and staff training outlined in local government playbooks (Durham local government AI pilot project playbook and practical steps); one memorable benchmark - track changes in average task turnaround time and frontline earnings exposure - to know quickly whether a pilot creates net community benefit.

Durham County leaders must plan for a fragmented U.S. regulatory landscape in 2025: no single federal AI statute exists, the White House issued an Executive Order in January 2025 signaling a deregulatory pivot while leaving agencies to apply existing statutes, and federal enforcers (FTC, EEOC, DOJ and others) will still use current consumer‑protection, civil‑rights and sectoral laws to police harmful AI uses - so expect enforcement even without new AI‑specific federal law (White & Case US AI regulatory tracker).

Simultaneously, state activity is surging: NCSL reports all 50 states introduced AI bills in 2025 and roughly 38 states enacted about 100 measures, including requirements for inventories, impact assessments, transparency and procurement rules that directly affect local government buying and deployment decisions (NCSL 2025 artificial intelligence legislation summary and analysis).

The practical takeaway: bake compliance into pilots and procurement now - use vendor clauses for bias audits, model provenance, and human‑in‑the‑loop controls - because only a small share of states had AI contract language ready in 2024, and failing to require these deliverables can slow approvals and create legal risk for HR, licensing, and high‑risk services.

Durham County can rely on the N.C. Department of Information Technology's practical playbook for safe AI adoption: the North Carolina State Government Responsible Use of Artificial Intelligence Framework (NCDIT) is a “living document” that lays out principles, practices and an enterprise risk‑management approach for all agency AI uses, and the specific guidance on Use of Publicly Available Generative AI defines enforceable safeguards - never enter PII or confidential case notes into public models, use state email for tool accounts, disable chat history for high‑risk prompts, and document yearly reassessments tied to the NIST AI RMF (NCDIT Guidance on Use of Publicly Available Generative AI) - so what this means for Durham is simple and immediate: treat any prompt containing client or service data as potentially public and design intake workflows to prevent data leakage.

NCDIT is also building workforce and governance capacity - appointing a statewide AI governance lead and publishing training and assessment templates - so align Durham's procurement, training, and audit checklists with state policy to accelerate approvals and reduce legal risk (NCDIT AI Governance and Policy Leader Announcement).

“The public has to put a certain amount of trust in the government,” said Cherie Givens, NCDIT Chief Privacy Officer.

Durham County's DCoBot rollout with Moveworks provides a compact, practical case study for other North Carolina counties: the Moveworks Copilot was integrated into Microsoft Teams and ServiceNow to serve 30+ departments and 2,200+ employees, automating routine IT requests so that within the first 30 days the bot handled a substantial volume of transactional inquiries and freed help‑desk staff to focus on higher‑value projects - a clear “so what” for procurement and workforce planners who need fast, measurable returns from pilots; the project stressed three operational priorities that other local governments should copy: continuous knowledge‑base curation for answer accuracy, tight infrastructure integrations with existing ITSM, and active adoption programs to build trust and usage (see the Durham County Moveworks case study and the broader market shift as ServiceNow moves to acquire Moveworks).

Early plans include expanding DCoBot to HR workflows to drive further administrative savings and employee self‑service.

“We're getting to a point where the bot is starting to do more of the transactional work, that includes providing step-by-step instructions to common IT inquiries or serving up forms for easy access and completion. Anything that's transactional, we're confident that the bot will be able to handle it.” - Zawadi Powell, IT Project Manager

Begin with a narrow, measurable pilot tied to a single pain point - start with IT or permitting workflows that have high volume and clear inputs, integrate the bot into existing channels (e.g., Teams or the county's ITSM) and run a 30–90 day sprint to prove value; follow a short vendor evaluation checklist (security, APIs, SLAs, exit clauses) and require a trial environment so stakeholders can test data handling and bias controls before full procurement - see a practical vendor evaluation framework for SaaS and migration tradeoffs for guidance.

Track one concrete metric from day one (for example, Mean Time to Resolution or turnaround time for a permit question) and aim for a measurable target - real cases show AI‑driven workflows can cut MTTR by up to 30%, which gives Durham a realistic “so what” to report back to commissioners.

Pair the pilot with a checklist for prompt/data classification (ban PII in public models, document human‑in‑the‑loop rules), designate owners in IT, procurement and HR for training and adoption, and publish a short after‑action report to inform scale decisions; for practical pilot steps and templates, see the Durham‑focused pilot playbook and vendor evaluation tips.

Durham local government AI pilot project playbook - practical steps for municipal AI pilots | SaaS vendor evaluation guide with MTTR examples and vendor checklist

Durham's procurement and governance playbook must turn known AI tradeoffs into contract terms: require independent bias and accuracy audits for high‑risk systems (for example, public safety gunshot‑detection analytics that raise accuracy, bias and oversight concerns), embed human‑in‑the‑loop controls and clear model provenance in RFPs, and fund workforce transition and reskilling for roles exposed to automation - records and permit‑processing staff are explicitly at risk and need concrete redeployment plans to avoid service gaps (public safety analytics tradeoffs in Durham, risks to records and permit processing staff in Durham).

Start every purchase with a narrow pilot and vendor trial that tests data handling, oversight workflows, and exit clauses - practical pilot templates help turn procurement into measurable outcomes, not open‑ended risk (practical AI pilot project steps for government procurement); the immediate “so what”: contracts that mandate audits and reskilling prevent a stalled deployment and protect service continuity when automation scales.

Durham's infrastructure decision is a choice among three practical tradeoffs: pure public cloud for rapid elasticity, hybrid consumption for balanced control, or on‑prem/private cloud for sensitive workloads - and recent HPE releases show clear paths for each.

For a hybrid-first strategy that preserves local control while adding cloud‑like consumption and AI operations, consider HPE GreenLake Intelligence: it converts GreenLake into an agentic, hybrid cloud that can run AIOps agents for networking, observability, FinOps and sustainability optimization (GreenLake Copilot beta expected Q3 2025), letting IT teams reduce manual toil without sending raw case data offsite (HPE GreenLake Intelligence hybrid cloud).

If Durham needs a turnkey, high‑performance private AI environment for training or regulated workloads, HPE's Private Cloud AI with NVIDIA offers a prevalidated private cloud and developer system (generally available Q2 2025) that keeps data on premises while accelerating model development (HPE Private Cloud AI with NVIDIA).

For municipalities worried about procurement and compliance, the practical “so what” is concrete: choose hybrid GreenLake to pilot high‑volume services with consumption pricing and agentic AIOps to surface cost and sustainability recommendations quickly, then move regulated or heavy AI training workloads to a Private Cloud AI node to retain sovereignty and meet local audit needs (see the government case for GreenLake in municipal/federal settings).

Start with a single workload - ITSM or permitting - and map it to the option above to measure MTTR, data residency, and total cost before scaling.

“To fully harness the power of AI, HPE and NVIDIA bring to market a comprehensive portfolio of AI solutions that accelerate the time to value for customers deploying generative, agentic and physical AI.” - Antonio Neri, President & CEO, HPE

Scaling AI across Durham County means measuring the right mix of technical and business KPIs, then using those signals to add new use cases where the numbers justify effort: track operational metrics (Mean Time to Resolution and first‑contact resolution), business impact (hours saved and cost avoidance), and adoption indicators (interaction volume and user satisfaction), as recommended in Acacia's KPIs guide for AI initiatives (Acacia KPIs guide for AI initiatives); Durham's DCoBot already proved the model by automating routine IT inquiries for 30+ departments and 2,200+ employees and handling a substantial volume of requests within the first 30 days, freeing service‑desk staff for higher‑value projects and creating a clear path to expand into HR and other transactional workflows documented in the Durham County Moveworks case study (Durham County Moveworks case study).

Use concrete, repeatable dashboards (latency, error rate, MTTR, hours saved, CSAT) and pilot phased rollouts so each new channel - HR, permitting, or citizen services - only scales when KPIs show real resource reallocation or time savings; a memorable benchmark: municipal Moveworks deployments have reported multi‑thousand hour annual savings and federal partners showed dramatic MTTR reductions, proving that measured scale can translate directly into staff capacity for strategic work.

"This transformation is not just about technology; it's about setting a new standard for public service in the digital age." - Chasity Locke, Assistant Director of IT Engagement and Communications

Next steps for Durham County are practical and sequential: first, align every pilot and procurement with the N.C. Department of Information Technology's Responsible Use framework and the GenAI operational rules (N.C. Department of Information Technology Responsible Use AI Framework) so contracts require bias and accuracy audits, model provenance, human-in-the-loop controls, and a ban on PII in prompts; second, run a narrow 30–90 day pilot tied to a single high-volume workflow (IT or permitting), measure Mean Time to Resolution and user satisfaction with a realistic 20–30% MTTR reduction target, then publish an after-action report to inform scale; and third, pair pilots with focused workforce training - use practical prompt and tool curricula such as the Nucamp AI Essentials for Work syllabus (Nucamp AI Essentials for Work syllabus and course details) - and share findings with state partners, following the precedent set by the NC Treasurer/OpenAI pilot which will publicly share learnings (North Carolina Treasurer and OpenAI public AI pilot announcement and learnings).

The clear payoff: a short, governed pilot that meets MTTR targets buys the credibility and measurable staff capacity needed to expand AI across Durham County services.

“Innovation, particularly around data and technology, will allow our department to deliver better results for North Carolina. I am grateful to our friends at OpenAI for partnering with us on this new endeavor, and I am excited to explore the possibilities ahead.” - Treasurer Brad Briner