People
- Prateek Shrivastava
- Gillous Harris
Over the past decade, enabling policies and the steady introduction of digital tools by providers have expanded financial account ownership to record levels. The new Global Findex 2025 reports that roughly 79 percent of adults now have an account — up from 51 percent in 2011 — yet meaningful use still lags for many people and places.
We are now tasked with translating access into value while the sector contemplates profound technological change — namely the adoption of generative AI (GenAI). Previous applications of AI were predictive. These systems process structured inputs to predict an outcome — for instance whether someone will repay a loan. GenAI moves beyond prediction to generate original outputs based on instructions received in natural language. GenAI can perform varied tasks such as report generation, conversational customer support, and even analysis of unstructured data. The technology is dynamic, human-like, and full of potential.
Why GenAI, and why now?
Generative AI (GenAI) can help close the gap between access and value by making products easier to understand, cheaper to deliver, and more responsive to context. At the same time, GenAI will be disruptive — changing the competitive landscape and business models of financial services providers. There will be winners, losers, and consequences that stakeholders are working to understand.
Most discussions of GenAI in the sector were speculative until recently; providers grappled with adoption and researchers lacked evidence. To move beyond speculation, the Center for Financial Inclusion (CFI) is authoring a series of articles informed by discussions with financial services providers, regulators, impact investors, and others that are actively working with GenAI. The series will attempt to shed light on common challenges, risks, and early success stories in GenAI adoption.
This first article offers a practical framework for where GenAI can add value across five levels of the ecosystem and draws on work and interviews with Accion Digital Transformation Fund, Accion Ventures, Santa Clara University’s Miller Center for Global Impact, Fundación Génesis Empresarial (Guatemala), FlexiLoans (India), and Banco Contactar (Colombia). If you are using AI in your work in financial inclusion, we’d love to speak with you as well. Please reach out to us at center@accion.org.
Early experiences reveal both promise and challenges, underscoring the need for an industry level discussion to help define what effective and responsible use of GenAI looks like for financial inclusion.
Based on interviews, as well as global evidence, the clearest benefits show up across five levels of the ecosystem: the client, the financial service provider, infrastructure operators, investors, and regulators.
GenAI’s most immediate benefit for clients is usability. Models can converse in local languages, explain terms in plain speech, respond to voice notes, and adapt to cultural context — no more one-size-fits-all interfaces. In Guatemala, Fundación Génesis Empresarial is piloting a messaging and voice assistant that explains loan options in Spanish and other local languages so rural borrowers can complete applications without a branch visit. In Colombia, Banco Contactar, the country’s largest private rural microfinance bank, is testing GenAI to automate collections with sensitive repayment reminders in local idioms. In Myanmar, ONOW is developing a white-label mentor that uses GenAI to coach women entrepreneurs and generate basic financial statements to help them access credit. And in India, Awaaz AI has deployed GenAI voice technologies to reach over 2 million microfinance and bank customers for onboarding, customer engagement, and transacting in regional languages. Used well, these assistants act as “explainers” — not just ticket-takers — nudging first-time users toward confident, sustained use.
For financial service providers, GenAI creates two big wins. First, it automates routine work so staff can spend time where judgment matters. IBM’s Digital KYC shows how GenAI can read policy, extract data from documents, and draft compliant narratives, shrinking onboarding from weeks to days. FlexiLoans is using GenAI to speed up product development and client onboarding processes, reducing time and cost to launch new services, which is especially useful in smaller Indian cities. Annapurna Finance is prototyping a human-in-the-loop GenAI copilot that produces plain-language explanations and “what if” repayment scenarios in multiple Indian languages; it also plans to integrate GenAI for policy reviews and risk reporting. And in Colombia, Banco Contactar is exploring the use of GenAI to extend tailored services across rural regions.
Second, GenAI makes personalization scalable. Digitally native firms can plug models into existing data pipelines to refine segmentation and underwriting; MFIs can start with copilots, automated document handling, and multilingual customer support, modernizing service without ripping out core systems. The net effect is faster decisions, lower marginal costs to reach last-mile clients, and products that better fit household realities.
The “plumbing” of inclusion — credit information, payment rails, and digital ID — can all benefit from GenAI’s strength with unstructured data. Credit bureaus and analytics providers can use unstructured or semi-structured telecom, utility, and transaction data to score thin-file customers more fairly. Payments operators can use a wider variety of data to detect anomalies in real time, improving safety for new digital users. KYC providers can generate precise, minimal document checklists that cut drop-off during onboarding. These behind-the-scenes gains matter: when onboarding is faster and safer, smaller loans and low-value accounts become economical to serve, widening the door for underserved clients.
For donors, DFIs, and impact investors, GenAI can speed up the work that determines where capital flows. Large models can summarize filings, harmonize social impact KPIs across investees, and surface early warning signals from portfolio reports. In practice, that means quicker reads on which MFIs or agri-lenders are expanding services to rural women responsibly, which pilots are stalling, and where technical assistance would change the slope. AI tools can help investors monitor competition, macro and micro economic factors, political changes, and peers of portfolio companies — all leading to better portfolio monitoring and risk analysis.
Regulators and supervisory are using AI to sift growing data volumes and draft risk insights from unstructured reports, complaints, and alerts. Authorities are deploying advanced tools for financial crime and fraud analytics; recent work emphasizes using GenAI to generate risk narratives and make data explorable in natural language. Namibia plans to deploy GenAI to intake, process, and analyze consumer complaints efficiently. For inclusion, these efforts matter: better suptech can spot emerging consumer harms faster and clarify expectations for safe, scaled deployments.
In short, GenAI is not about more products, but about increasing efficiency, driving greater reach, and innovating for better engagement—across the inclusion stack—helping financial services become more affordable, understandable, and responsive to client realities.
Not just potential — also new risks
GenAI can deepen inclusion, but it also raises risks — especially in thinly regulated settings and low-literacy contexts. While we lack the space to discuss them thoroughly in this article, these risks include:
- Discrimination: Integrating GenAI into underwriting or fraud controls can amplify bias and reduce transparency. Consumer-facing assistants can also produce toxic or misleading outputs without careful guardrails, particularly in local languages with sparse training data. For more on algorithmic bias in inclusive finance, see CFI’s report here.
- Privacy: Models introduce new attack surfaces and can inadvertently leak sensitive data. Clients may overshare with chatbots unless systems are designed to minimize and protect data. To learn more about privacy by design, read CFI’s guide here.
- Reliability: Variability is part of what makes GenAI feel human — but it can also produce errors. When advice is wrong or a model mishandles a case, consumers bear the cost unless escalation paths and redress mechanisms are clear and effective.
The answer is not to slow innovation to a crawl but to embed fairness, transparency, and client protection into every deployment — and to measure client understanding and outcomes, not just speed and cost. Standards are urgently needed to spur adoption of effective safeguards. At the same time, we need to innovate and build evidence about what GenAI can (and cannot) do for consumers.
So, where should we start?
CFI sees three immediate steps for the sector:
- Build evidence. Invest in research and pilots to test where GenAI works, where it fails, and what guardrails are required.
- Define standards. Convene providers, regulators, funders, and innovators to set responsible AI norms for inclusive finance.
- Support problem-driven innovation. Back teams that are willing to experiment, be transparent with results, and share lessons learned.
GenAI will not, on its own, fix structural barriers or replace the human relationships that underpin trust. But it is likely to be one of the most consequential tools of this decade. By engaging now, we can steer GenAI toward empowerment rather than exclusion.
Are you using AI in your work in financial inclusion? We would love to hear from you! To share your experience and be considered for this series on the use of AI in our industry, please contact us at center@accion.org. We look forward to hearing from you.
Authors
Explore More
