Although micro, small, and medium enterprises (MSMEs) form the backbone of both rural and urban economies in Ghana, their owners face persistent barriers to financial inclusion. Despite representing the vast majority of the country’s businesses, employment, and GDP, many MSMEs, particularly those in rural areas and those owned by women, operate informally, have limited access to financial and literacy skills training, and must often resort to high-interest instant digital loans or informal borrowing since traditional financial services remain out of reach.
While still in early stages, the burgeoning field of artificial intelligence (AI) offers new possibilities to help close the gap between MSMEs and formal financial service providers (FSPs), thereby enabling small business owners to expand their operations, enhance their productivity, and thrive.
In Ghana, the UN Capital Development Fund (UNCDF) is leading the Leveraging Digital Ecosystems for Increased MSMEs' productivity project, a Joint SDG-Fund-financed Joint Programme (JP) implemented together with the UN Development Programme (UNDP) and UN Conference on Trade (UNCTAD). Among other objectives, UNCDF is exploring how the power of AI is already being harnessed to benefit MSMEs in the country and how the agency can leverage its unique role to catalyze and de-risk investment in innovative, AI-driven solutions to bridge longstanding gaps in access, affordability, and trust.
This article represents a first look into existing AI solutions on the supply (FSP) and demand (MSME) sides of the market and how UNCDF can unlock its potential.
Supply-side innovation: AI for financial service providers
From digital lenders to microfinance institutions and from lending and savings providers to mobile money providers, several FSPs in Ghana have already embraced AI and machine learning tools that are enabling them to expand their clientele and serve smaller businesses.
Fido Credit, a leading digital lending platform and UNCDF partner under the JP, has developed a wide range of proprietary AI and machine learning models including document validation and classification, fraud detection systems, vision-based Know Your Customer (KYC) tools, and financial forecasting. According to Noa Lubin, Fido’s Director of Data Science, these tools, which are embedded in both Fido’s onboarding and ongoing customer engagement processes, are enabling it to scale its lending without heavily relying on manual underwriting, improve repayment forecasting, make faster and more accurate decisions, and reduce fraud losses and operational costs. Importantly, these AI-driven tools are also allowing Fido to serve a broader range of customers, including traditionally underserved microenterprises that too often face challenges such as a lack of or limited documentation and formal credit history, among others.
Credit scoring
AI is creating a fundamental shift from traditional lending approaches. While lenders have historically required extensive collateral and formal financial histories, which most MSMEs lack, AI-powered platforms like Fido’s use risk acquisition, behavioral, and recovery models to evaluate mobile money transaction patterns, business location data, and other digital footprints to make more informed lending decisions.
These new tools are also allowing FSPs to extend credit to smaller and less formal businesses. For example, Fido’s AI-powered credit scoring tools, including its Fido Score Ecosystem, have enabled it to lend to micro and small businesses that would not qualify for loans under traditional credit systems. By drawing on alternative data to evaluate MSMEs’ credit worthiness, Fido’s AI-driven tools provide fairer evaluations to business owners and have given Fido confidence to offer credit without the need for traditional collateral. According to Noa Lubin, MSMEs are also benefitting from these tools in the form of faster loan approvals and more tailored credit offers.
While the use of alternative data in AI credit scoring is promising, it still comes with limitations and challenges. For example, although alternative data is enabling FSPs to reach historically underserved markets, many informal MSMEs still do not generate sufficient digital footprints to be rated by these systems. When potential customers do generate sufficient data, this information is often subject to regulations and privacy concerns, with strict consent and fairness requirements limiting the extent to which it can be leveraged by FSPs. Furthermore, alternative data is not always clean and easy for modelers to work with. As one industry practitioner explained, “alternative data is often noisy, inconsistent, and difficult to standardize compared to traditional financial records.”
Fraud prevention and risk management
AI's impact extends beyond credit decisions to fraud prevention. For instance, Fido uses anomaly detection machine learning models that analyze transaction patterns to identify suspicious activity. Dr. William Derban, the Head of Programs and Partnerships in Digital Innovation at Opportunity International whose Ghanaian subsidiary (Opportunity International Savings and Loans, OISL) is a UNCDF partner on this JP, suggested that an additional layer of fraud detection could take the form of an AI chat bot that loan officers can query. Intelligent systems like these allow FSPs to balance security with accessibility, protecting customers without excluding legitimate users. As another industry expert noted, the challenge lies in “evolving fraud tactics and balancing strict controls with the customer experience.”
Additional AI opportunities for FSPs
AI is opening up scores of new possibilities for improving analytic accuracy, lowering risk and costs, and improving operational efficiency for FSPs which can, in turn, allow them to expand their client base to smaller enterprises. In a recent stakeholder engagement exercise with UNCDF, industry experts from the Ghanaian ecosystem suggested novel ideas for using AI to support supply-side FSPs. These include predictive tools for small business cash flow forecasting, leveraging more of the data collected on loan application forms into the decision-making process, and synthesizing financial and non-financial information. There is also significant potential in leveraging conversational large language models (LLMS, or chat bots) for various purposes, including to help microenterprises navigate complex and cumbersome credit applications and other forms, to provide customer service, to support dispute resolution, and to assist loan officers.
Demand-side potential: AI for MSMEs
While FSPs are quickly adopting AI applications that are benefitting MSMEs, there is significant opportunity to create new AI-driven tools that support MSME owners directly. Although many of the lowest-income MSME owners still rely on basic and feature phones, and much work remains to close the digital divide, smartphone ownership is rapidly increasing across Ghana.[1] According to one local entrepreneur interviewed for this piece, entrepreneurial business owners and young people are particularly keen to get a feel for how AI works and how it can benefit them.
Record-keeping and analytics
In its effort to provide solutions that are user-centered, UNCDF’s Ghana team recently caught up with Dennis Gyamfi, a bank and mobile money agent in the Ashanti region of Ghana, to understand his perspective, learn from his experience, and envision potential uses for AI for MSMEs. Dennis maintains his business records in MS Excel and, at his own initiative, has recently begun using Gemini, ChatGPT, and Copilot to support his analysis. He uses these tools to check data quality, provide Excel formulas to enable him to calculate key business metrics, and output performance metrics based on simple descriptions. “It has been very, very useful,” he explained. “[AI] is the only system that can give you information that is helpful like this without cost…. This is the type of advice that I would have engaged experts for.”
While Dennis has not spoken with other financial service agents or his MSME customers about their use of AI tools, he believes that they would benefit from using AI for similar purposes. He suggested that AI could be useful in developing a single access point through which he and other bank and mobile money providers could access numerous different custom platforms each day. Dennis also raised the idea of using AI tools to track customers’ expectations, experiences, and interactions to enable agents to better understand how to relate to different customer segments and better sustain their client relationships over time.
Not all MSME owners in Ghana keep the same level of records as Dennis, but AI still has the power to transform their businesses. A recent UNCDF field visit in Takoradi market in western Ghana provided an important reminder that some microenterprises do not keep even basic records. These business owners could benefit from simple digital daily sales and expense tracking systems that could generate business records to support MSME owners’ decision-making and easily feed into loan applications. Integrating AI into these tools could revolutionize the businesses of MSME owners who are not literate or numerate. For instance, conversational and visual interfaces could allow users to speak in local languages, upload photos of handwritten records or stock, and receive reports and insights through audio or simple graphics.
AI chatbots for better business decisions
Dr. Derban further shared some of his ideas for demand-side AI solutions. Drawing from Opportunity International's successful agricultural AI chatbot – which provides farmers with timely, authoritative information through WhatsApp – he suggested an AI tool that crowdsources information from a trusted network could deliver market intelligence, seasonal business guidance, and supplier connections to microenterprises in an easily accessible way.
The key insight from the human-centered design (HCD) process that informed Opportunity International’s chatbot applies directly to financial services: people often want timely information more than just credit because knowledge enables better business decisions, including more informed and more confident borrowing. As Dr. Derban explained, “I thought they were going to ask about how they could get more money, but they didn’t…. They asked for timely information. Things are changing quickly. [If I’m a farmer and] I know a little bit about the future then … I know what to plant, I can assess my profitability, I can take a loan and have some assurance that I will be able to repay it.”
AI chatbots that are trained on vetted digital and financial literacy content and local languages offer unique opportunities to provide MSME owners with digital and financial education. As Dr. Derban explained, “the beauty of large language models is that you can do this in a conversational form. You are engaging in the way you would usually do, and you don’t have to be educated to use it. You don’t need to speak English or be able to write. Now, you can speak in your local language and still get the same level of service.”
Beyond standard digital and financial literacy training, AI chatbots could offer MSME owners personalized business advice and decision support based on their individual circumstances and local market conditions. , a prospect that Dr. Derban finds exciting and Dennis enthused, “would be one of the best tools ever.” Such a tool could leverage local and sectoral business data to help MSME owners make the most of seasonal trends, for instance deciding what products to sell or what activities to engage in at different times of the year, provide tailored performance reviews and recommendations, and more.
Among the many potential use cases, such a tool could also support businesses in deciding when and whether to seek a loan. As Dennis explained, given high interest rates, this is a critical decision that can make or break a business. Many MSME owners do not fully understand how to make this decision and what their other options are. For instance, he shared that some business owners think they need more money when what they really need are connections to a new supplier. While some microentrepreneurs may have a plan to repay a loan quickly, for instance using credit to cover short-term dips between busy seasons, others may not be able to expect a sufficient surge in profit that will enable them to repay both the principal and accumulating interest, which can lead them to take out secondary loans to repay the first. An AI decision-support tool could therefore help users understand when to seek a formal loan, where they can access credit with the best terms, how much credit they really need and how to use that credit most effectively, and repay it.
Consumer fraud protection
While FSPs focus on institutional fraud prevention, consumers need protection from rampant scams that threaten to undermine their trust in digital and financial services. Chatbots could make fraud-related educational content more interactive or allow users to text in questions like, “I just received this message: [...] Is it a scam?” AI-driven tools employed by telecommunication companies or embedded in phone software or messaging apps like Whatsapp could provide next-level fraud protection, for instance by filtering calls from numbers that crowdsourced data identify as likely scams, helping users assess the likelihood that a message or phone call is fraudulent, and offering real-time alerts about suspicious communications.
Strategic investment opportunities
A plethora of opportunities exist to leverage AI to improve digital financial inclusion for MSMEs on both the supply and demand sides. Perhaps the most powerful potential lies in the interconnections between supply and demand. For example, if FSPs develop easy-to-use, AI-powered MSME record-keeping and business intelligence tools, the FSPs can, in turn, leverage this data to improve the accuracy of their credit scoring and risk models and increase their confidence in lending to historically underserved groups, thus boosting their market reach. In turn, more accessible and affordable credit can enable business growth that makes MSME entrepreneurs better customers. Fraud protection on both sides builds ecosystem-wide trust in digital financial services.
This creates a virtuous cycle where improvements on one side amplify benefits on the other. As Dr. Derban observed, "if we reduce the challenges [that MSME owners] face, they can grow and be profitable, and then it's easier for the banks to engage with them."
Rather than viewing supply and demand-side solutions as separate initiatives, development partners can leverage the whole ecosystem to create integrated approaches that multiply impact.
Catalytic actors such as UNCDF are in the position to further enable Ghana’s AI ecosystem by contributing to the development of supportive regulatory frameworks, facilitating broader industry collaboration, and injecting and catalyzing funding to test innovative, human-centered pilots. Such investments can help overcome persistent challenges of evolving regulations, high implementation costs, limited data-sharing infrastructure, and data privacy concerns.
Under this JP, UNCDF is strategically positioned to fill market gaps where private sector incentives alone may be insufficient. With its unique mandate, UNCDF is leveraging its capability to de-risk early-stage investments of partner FSPs to finally help them to crowd-in finance to scale up development impact. Through technical assistance and investment grants, UNCDF is supporting financial intermediaries to design, test and pilot financial innovations to better serve vulnerable groups that otherwise would remain excluded. These include translation and voice technologies for local languages, fraud prevention tools focused on consumer protection rather than just institutional risk, easy-to-use business management tools designed for informal MSMEs, and content development that ensures AI-powered training tools provide authoritative, locally relevant guidance.
AI is unlocking unprecedented opportunities to leverage technology for inclusive development. The path forward requires continued collaboration between the private sector to push the boundaries, the public sector to provide the necessary regulatory environment, and risk-taking actors such as UNCDF to demonstrate financial viability where markets fail, build pipelines of investable opportunities, and create the conditions for additional development finance and private capital to follow.
[1] According to the National Communications Authority, as of Q1 2025, 16 percent of the country’s 40.5 million devices were basic phones, 26 percent were feature phones, and 58 percent were smartphones. The number of smartphone devices – 23.7 million – was up from 18.7 million as of Q4 2023, just over one year prior.
Technical writing: Jennifer Spencer
Co-author: Yaa Owusu-Amoah
Note:
All joint programmes of the Joint SDG Fund are led by UN Resident Coordinators and implemented by the agencies, funds and programmes of the United Nations development system. With sincere appreciation for the contributions from the European Union and Governments of Belgium, Denmark, Germany, Ireland, Italy, Luxembourg, Monaco, The Netherlands, Norway, Poland, Portugal, Republic of Korea, Saudi Arabia, Spain, Sweden, Switzerland for a transformative movement towards achieving the SDGs by 2030.
