The Complete Guide to Using AI in the Financial Services Industry in Charlotte in 2025

Charlotte's financial hub is already running AI at scale: Bank of America reports more than 90% employee adoption of internal AI tools, coding assistance that improved developer efficiency by over 20%, and automation that frees “tens of thousands of hours” for client engagement - concrete examples of how local banks turn models into measurable productivity and faster service (Bank of America 2025 AI adoption report).

That same drive toward efficiency must be balanced with oversight - AI in credit decisions and customer-facing systems raises regulatory and compliance questions that Charlotte firms should plan for early (AI and financial services regulatory compliance review).

Practical workforce readiness matters: teams that learn prompt engineering and low-code AI workflows reduce rollout risk and accelerate results (Nucamp AI Essentials for Work syllabus (15-week program)).

“AI is having a transformative effect on employee efficiency and operational excellence.” - Aditya Bhasin, Chief Technology & Information Officer, Bank of America

Charlotte's AI ecosystem is pragmatic and fast-moving: large banks have moved from pilots to scale (Bank of America reports Erica serving roughly 20 million clients and more than 2.5 billion interactions, with AI embedded across employee tools and developer workflows), regional banks like Truist and Wells Fargo operate production virtual assistants for hundreds of common inquiries, and a rising roster of startups and spin‑outs - including a recent local AI company founded by former Truist and Credit Karma executives - are building niche agents and automation for finance and operations; at the same time, civic and academic institutions (UNC Charlotte's AI4Health Center and TAIMingAI) are closing the talent and governance gap so Charlotte can adopt responsibly rather than reactively.

The so‑what: these combined forces mean Charlotte is not just running chatbots, it's testing governance, workforce reskilling, and customer‑experience models at scale - lessons other regional financial hubs will watch closely (Bank of America 2025 AI adoption report on Erica and enterprise AI, UNC Charlotte AI centers and Charlotte citywide AI initiatives).

“AI only goes so far right now,” says Herman Nicholson, “but give it a couple of years.”

Charlotte financial firms should prioritize two parallel AI tracks: customer-facing assistants that improve engagement and internal automation that frees human expertise for revenue work.

On the client side, Bank of America's Erica demonstrates scale - about 20 million clients and more than 2.5 billion interactions - where proactive insights (roughly 60% of usage) surface timely suggestions like subscription monitoring and balance alerts that reduce simple inquiries and keep customers informed (Bank of America 2025 AI adoption report on Erica, Charlotte Magazine article on AI in Charlotte and Erica).

Internally, widespread employee AI (over 90% adoption for internal assistants) plus GenAI-powered coding tools and automated client-meeting drafts have delivered concrete throughput gains - coding efficiency >20% and “tens of thousands of hours” reclaimed for client engagement - so the measurable so‑what is simple: automated triage and drafting converts hours of routine work into higher-value advisor time.

High-impact pilots to run first in Charlotte: proactive alerts and dispute triage in contact centers, KYC/onboarding automation to cut time-to-account-open, and AI summarization of research and call recordings to speed decisions while preserving human oversight.

“AI is having a transformative effect on employee efficiency and operational excellence.” - Aditya Bhasin, Chief Technology & Information Officer, Bank of America

Charlotte financial institutions should treat the Section 1033 / Personal Financial Data Rights regime as an operational priority even while the CFPB has paused litigation and signaled new rulemaking: the October 2024 rule already prescribes machine‑readable data, separate consumer and developer interfaces, LEI identification, purpose‑limited uses by authorized third parties, and GLBA/FTC‑aligned information‑security programs - and it establishes phased compliance dates that start April 1, 2026 for the largest providers, so API readiness, vendor contracts, and data‑mapping must move from “future project” to near‑term deliverable (CFPB Personal Financial Data Rights rule standards and obligations (October 2024); CFPB stayed the Section 1033 open banking rule and initiated new rulemaking).

Practical, Charlotte‑focused actions: run a covered‑data inventory, map flows to third parties, plan LEI acquisition, upgrade security programs to meet GLBA/FTC expectations, and engage industry standard‑setters so local APIs meet performance and consent requirements - doing this now turns regulatory uncertainty into a competitive edge for banks and fintechs that can offer compliant data portability and faster onboarding (DLA Piper implementation takeaways and phased deadlines for Section 1033 compliance).

The so‑what: teams ready by the first compliance dates can reduce integration risk and win customer trust while competitors retrofit governance under pressure.

“the limited legislative history confirms what the statute's text and structure make clear: the statute was intended simply to ensure that consumers would have access to their own information.”

Charlotte firms deploying AI should bake human-in-the-loop controls and clear ethical disclosure into every high‑risk workflow: require a human reviewer for adverse‑action outcomes (credit denials, price‑sensitive trading alerts, or regulatory escalations), log model rationale for audits, and publish customer‑facing disclosures that explain when AI helped make a decision and how to request human review - practices grounded in both industry examples (nCino's Banking Advisor keeps a “human in the loop” for compliance and oversight) and regulatory guidance that treats AI outputs as decision‑support rather than replacements for human judgment (nCino nSight 2024: Banking Advisor; RGP: AI in Financial Services 2025).

Aligning escalation rules to a risk‑based “sliding scale” (higher explainability and mandatory human review for credit scoring, fraud, or trading) and keeping auditable model logs will meet emerging expectations from supervisors and help Charlotte teams turn explainability into a competitive trust signal (GAO: Use and Oversight in Financial Services).

“A human-in-the-loop approach is a must to mitigate the accountability risks posed by AI.”

Charlotte financial institutions choosing between internal platforms and public models should take a risk‑first, value‑focused approach: use private, on‑prem or VPC‑isolated models for regulated, high‑risk workflows (credit scoring, KYC/onboarding, payments) to preserve data control and auditability, while leveraging public frontier models for low‑risk productivity tasks and rapid prototyping - stitched together with retrieval‑augmented generation (RAG) to keep proprietary records private.

The Bank of England highlights concrete hazards to weigh - prompt‑injection, data‑poisoning and vendor concentration that can turn a single cloud outage into a city‑wide service disruption - so operational resilience and exit planning must be built into vendor contracts (Bank of England guidance on AI in the financial system - April 2025).

Infrastructure and connectivity choices matter: secure API pipelines, zero‑trust auth, and nearby interconnection reduce latency and exposure when hybrid architectures call public models for reasoning but keep sensitive inference and logging inside private data fabric (Digital Realty article on scaling private AI with public models).

Finally, follow the practical playbook in PwC's 2025 guidance: prioritize AI strategy, proprietary data integration and phased pilots over debating which LLM to pick now - because embedding institutional data safely into models is what actually moves the needle for risk‑adjusted value.

Start with small, measurable pilots that convert routine work into clear ROI: (1) KYC and account‑onboarding automation to shorten time‑to‑account-open and reduce manual errors (pair OCR/NLP document extraction with human review for high‑risk fields); (2) AI document‑automation for loans, tax forms and compliance reports to slash processing time and errors (case studies show up to 20x faster approvals and large cost cuts when document AI handles diverse file types); (3) contact‑center triage that uses retrieval‑augmented generation (RAG) and automated dispute routing to remove repetitive tickets and free advisors for revenue work; and (4) community partnership pilots that address the local digital divide - integrating workforce upskilling and device access so customers can complete remote verification and digital onboarding.

Tie pilots to local assets: the City of Charlotte's Access Charlotte expansion serves up to 8,600 households across 75+ sites and embeds digital navigators and device programs, which makes a workforce‑readiness pilot practical at scale (Access Charlotte expansion program and digital navigator initiative).

Learn from public‑sector experimentation too: North Carolina's 12‑week OpenAI pilot with the State Treasurer explored AI for operational efficiency and analyzing public financial data - lessons that financial firms can adapt for safe, auditable prototyping (North Carolina 12-week OpenAI pilot for state government efficiency).

For technology selection, validate accuracy, integration effort and human‑in‑the‑loop checkpoints during a 60–90 day proof‑of‑concept, then measure time‑saved per FTE and error‑rate reduction before scaling; documented document‑automation wins provide a concrete “so what”: proven speed and cost improvements that underwrite broader AI rollout (AI-powered document automation for financial services guide).

“Internet access today is foundational for learning, working, interacting with the government and other family members. It's foundational to everyday life.”

Charlotte's talent pipeline for AI in financial services centers on UNC Charlotte's industry‑aligned bootcamps: fully online, monthly start dates and small cohorts (maximum five students) make rapid upskilling practical for working professionals and local hires, while a hands‑on curriculum (Python, cloud computing, ML, NLP, LLMs) plus a capstone that delivers three portfolio projects gives employers verifiable, work‑ready artifacts and career‑services support for placement (UNC Charlotte Artificial Intelligence Bootcamp - program details); the university's broader bootcamp catalog offers parallel tracks (data science, cybersecurity, software engineering) that firms can tap to reskill staff quickly (UNC Charlotte Professional Development Bootcamps - catalog and enrollment).

The practical so‑what: local banks and fintechs can recruit or upskill workers who have completed project‑based LLM and ML training on predictable schedules, reducing time to pilot staffing and helping teams move from proof‑of‑concept to production with fewer hiring gaps.

Program: Artificial Intelligence Bootcamp (UNC Charlotte, powered by Flatiron School)
Delivery: 100% online; monthly start dates (first Monday of each month)
Duration: Full‑time: 12 weeks; Part‑time: 36 weeks
Cost: $9,900 (one‑time); loan and payment options available
Class size: Maximum of five students
Capstone: Three portfolio projects (regression, supervised classification, unsupervised model)
Contact: professional@charlotte.edu • 704‑687‑8900

Charlotte financial firms adopting AI must treat cyber risk, fraud defense, and total cost of ownership as a single program - not separate line items - because the cost of a successful compromise can be immediate and existential (PNC's fraud examples include payment‑redirection schemes and construction‑contract impersonations that have produced losses in excess of $1 million).

Practical defenses start with the basics PNC prescribes - formal, documented cybersecurity programs, annual risk assessments, strong access controls, encryption in transit and at rest, third‑party assurance and SOC reporting, secure SDLC practices, and tested incident‑response and business continuity plans - and extend into AI‑specific controls like adversarial‑simulation testing and identity‑proofing to counter AI‑generated stolen or synthetic identities highlighted in upcoming industry forums.

Build TCO models that include recurring line items (annual third‑party audits and SOC reports; continuous monitoring and threat intelligence; periodic tabletop and adversarial exercises; and recurring training for phishing and payments‑fraud scenarios) and compare those costs to measured fraud losses and recovery expenses; doing so turns security from a compliance checkbox into a quantifiable risk‑management lever.

For practical peer learning and threat updates, attend regional gatherings where practitioners exchange playbooks and countermeasures - Datos Insights and local FinCrime forums convene sessions on AI risk and identity fraud - and use those insights to prioritize vendor assurance, human‑in‑the‑loop checkpoints, and encryption/segregation strategies during procurement and pilot planning.

PNC Cybersecurity Resource Guide for Financial Institutions • Datos FinCrime & Cyber Security Forum: Deduce Session on AI and Identity Fraud

Getting started in Charlotte means pairing short, measurable pilots with immediate security and governance work so experiments don't become compliance headaches: begin with a covered‑data inventory and vendor contract review, require centrally managed authentication and multi‑factor authentication for any system that handles sensitive financial data, and map data classification rules to your pilot environments so test systems use the same protections as production - practical steps drawn from the UNC Charlotte Information Security Checklist will make your pilots auditable and resilient (UNC Charlotte OneIT Information Security Checklist).

Staff readiness is equally important: enroll operations and product owners in a structured, workplace‑focused program (for example, the Nucamp AI Essentials for Work 15-week syllabus) so your teams can run prompt‑engineered proofs of concept with human‑in‑the‑loop controls and measurable KPIs (time‑saved per FTE, error‑rate reduction) before scaling.

The so‑what: projects that start with these three controls (data inventory, MFA/access controls, vendor assurance) move faster through procurement and reduce regulatory friction, turning early AI pilots in Charlotte into repeatable, bankable services.