AI With Returns, Not Hype: Vodacom Business Is Turning Telco Strengths Into FinTech Outcomes
In this episode, “AI With Returns, Not Hype: Vodacom Business Is Turning Telco Strengths Into FinTech Outcomes”, of Talking Success, The Best Fintech Podcast, Darren Franks sat down with Kevin Odudoh, Ati Ngubevana and Peter Malebye from Vodacom Business. They tell us how they are approaching AI with a clear business lens, drawing on capabilities proven inside the Group and now offered to enterprise clients. The focus is practical: financial inclusion, fraud reduction, compliance and measurable ROI.
AI is on every executive agenda. The tough bit is not the technology itself. It is proving that the investment will lift revenue, protect margins and strengthen trust. Below is a distillation of the first half of that conversation, shaped for founders, product leaders and risk executives across African financial services. If you want AI to do more than generate demos, this is for you.
Start with the why, not the model
Vodacom Group’s AI and automation programme is guided by three aims: enable growth, create simplicity and build profitable, scalable capabilities. That is the filter used to decide which models to build, which problems to tackle and which solutions are worth commercialising. In other words, the business case comes first. The model choice is a consequence, not the headline.
This internal discipline matters when you take AI to market. Clients want to see evidence that the thing you are selling has already created value inside your own organisation. Vodacom’s team stresses that they are not pitching slideware. They are taking capabilities that have been deployed and battle-tested within the Group and packaging them for enterprise use.
Telco data as a unique advantage
Telcos sit on rich, continuously updated behavioural data. With appropriate controls, that data can be transformed into features that are predictive of financial risk and opportunity. This is where a telco can bring something different to a bank or a FinTech.
Two examples from the Vodacom playbook illustrate this advantage.
1) Behavioural credit scoring that expands inclusion
Traditional bureau scores exclude millions of people across African markets. By combining signals from telco networks and, where relevant, mobile money activity, Vodacom has developed a behavioural credit score that helps lenders see beyond thin-file limitations. The result is access. Lenders can responsibly extend small-ticket finance or device financing to customers who previously sat outside the system.
The goal is not to replace existing risk models. It is to add a supplementary lens that improves approval rates without lowering standards. Done right, that unlocks top-line growth for lenders while supporting financial inclusion in a way that aligns with national objectives.
2) Proactive fraud detection that sees the network
Fraud is a networked problem. Telco graphs reveal that network. Vodacom has built a proactive fraud identification capability that looks for behavioural patterns associated with confirmed fraud and then maps the connections around those numbers to surface likely syndicated behaviour. This was built to protect Vodacom’s own business. The next step is to expose it as an API to financial institutions at critical points in the customer journey.
Picture an onboarding or loan application flow. A mobile number is provided. An API call checks whether that number exhibits attributes previously linked to fraud in the telco environment. The lender receives a risk signal in real time and can step up verification or decline. That protects the loan book, reduces collections pressure and saves money that would otherwise be lost to bad actors.
Taken together, these two capabilities speak directly to what every Exco wants from AI. The first helps you grow. The second helps you keep what you grow.
Compliance, consent and doing AI the responsible way
Great data brings great responsibility. The team is frank about the governance challenge. Innovation must live within the rules, and those rules are evolving. The approach covers a few key principles:
- Anonymisation and aggregation by default. Insights are derived without exposing personally identifiable information unnecessarily.
- Consent where it counts. Use cases that require linkage to an individual operate with explicit consent and transparent terms.
- Use within a governed platform. Data is ingested, stored and processed on platforms that meet privacy and security expectations for regulated industries.
- Independent assurance. Where appropriate, external partners and cloud providers are selected and configured to satisfy regulatory scrutiny.
A helpful historical example is the COVID collaboration with government, where anonymised mobility data was used to identify hotspots. The point is not the pandemic. It is the pattern. Vodacom has experience turning sensitive datasets into public-interest insights without compromising individual privacy. The same mindset now guides financial services use cases.
The neutral orchestrator in a crowded ecosystem
If you are a FinTech founder, a bank CIO or an insurer’s head of data, you do not need another vendor forcing you onto a closed stack. You need a partner that manages connectivity, platforms and data flows, helps you plug in best-of-breed components and keeps you onside with regulators. Vodacom Business describes its role as a neutral orchestrator.
Practically, that means:
- Data rails and landing zones. Secure ingestion from multiple sources, governed storage, clear lineage and robust access control.
- Interoperable platforms. The ability to run workloads and serve APIs in environments that meet industry requirements rather than locking you into one tool.
- Open integration. FinTechs, banks and insurers can bring their own models, augment with Vodacom-exposed signals and build products that fit their markets.
- Operational resilience. Telco-grade reliability for services that sit in the critical path of payments, onboarding and lending.
For financial services leaders, this orchestration layer reduces the friction of moving from pilot to production. It also clarifies accountability. Someone is responsible for keeping the pipes clean, the platforms secure and the lights on.
Where the ROI lands in financial services
Executives understandably ask about return on investment. Here is how the above translates into numbers.
- Approval uplift without higher losses. Supplementary behavioural scores increase acceptance rates for targeted segments. If your baseline approval is 30 percent and you lift that by five to ten points in a controlled way, the revenue impact is immediate.
- Fraud loss avoidance. Stopping a fraudulent application before disbursement has a double benefit. You avoid write-offs and you avoid downstream operational costs in collections and dispute handling.
- Lower cost to serve. AI-assisted decisioning and process automation reduce handle time in onboarding, KYC and customer support.
- Better pricing. Usage-based insurance and risk-aligned lending allow you to price closer to true risk, which improves margin and competitiveness.
- Trust and retention. Customers who feel protected from fraud and receive fair, fast decisions are more likely to stay, spend and recommend.
These gains accrue when AI is embedded into business processes rather than run as a side project. Which is why the orchestration layer and responsible data practices are not optional extras. They are prerequisites.
Data sovereignty and the “no ChatGPT on our Wi-Fi” moment
Many organisations have reacted to generative AI by blocking public tools on corporate networks. The fear is simple. Sensitive data might be pasted into a public model and leak. That concern is legitimate and has been amplified by stories of employees feeding confidential material into consumer apps.
A useful way forward is to separate three ideas:
- Foundation models. These can be hosted privately or accessed via enterprise contracts that disable training on your prompts and outputs.
- Domain signals and features. This is your proprietary advantage. In finance, that includes transaction graphs, device behaviour, credit outcomes and telco attributes. Keep them governed and close.
- Decisioning and workflow. This is where models meet process. Keep it inside your perimeter, log everything and design for human oversight.
Vodacom’s stance aligns with this separation. Use private or enterprise-grade models where appropriate. Keep sensitive signals in a governed platform. Expose only the minimum needed via APIs with clear contracts. That allows companies to benefit from generative interfaces and task-specific models without compromising data sovereignty.
Practical applications you can deploy now
If you are mapping your 2026 roadmap, these are the near-term, high-impact applications suggested by the team’s work:
- Onboarding risk checks. Enrich applications with telco-derived fraud risk signals to flag high-risk numbers before KYC is completed.
- Thin-file lending. Add a behavioural telco score as a supplementary feature in your credit models to increase approvals in segments previously excluded.
- Device and airtime finance. Use a combined telco and payments view to manage risk and expand access to essential connectivity.
- Usage-based insurance. Leverage IoT connectivity and data flows to support fairer, behaviour-linked pricing without exposing PII.
- Customer support triage. Apply AI to classify intent, recommend next actions and surface fraud signals to agents in real time.
- Collections prioritisation. Use network behaviour changes as early warning indicators to prioritise outreach before delinquency hardens.
Each of these can be piloted quickly, monitored closely and scaled once the business case is proven.
Guardrails that keep you out of trouble
No AI rollout is risk-free. Here are the guardrails discussed that help you move fast without breaking trust:
- Model governance. Document purpose, training data, performance, bias checks and monitoring. Treat models as living assets.
- Privacy-by-design. Build anonymisation and minimisation into the data pipeline. Make consent meaningful, not a tick-box.
- Security posture. Assume that fraudsters will probe your interfaces. Harden APIs, rate-limit sensibly and audit access.
- Human-in-the-loop for key decisions. Keep skilled people in the chain where the cost of a wrong decision is high.
- Transparent customer communication. If AI influences a decision, be ready to explain how, in plain language.
- Regulator engagement. Share your approach early. Co-create guidelines rather than waiting for rules to arrive.
These steps do not slow you down. They save you from expensive rework and reputational damage.
Partnership over procurement
Perhaps the most important message is philosophical. Vodacom Business is not trying to be your everything. It is offering to be the secure rails, the neutral platform and a set of proven capabilities that you can compose with your own. That leaves space for FinTech creativity, bank-grade controls and the reality that different markets have different needs.
It is also a partnership that runs deep in the ecosystem. Vodacom’s support for platforms like the African FinTech Festival and the South African FinTech Awards reflects a belief that inclusion and innovation are team sports. Telcos, banks, FinTechs, insurers, regulators and cloud providers all have roles to play. Orchestration makes the music.
The takeaway for leaders
If you remember one thing, make it this. AI is not a magic wand. It is an amplifier. In financial services it should amplify inclusion, fraud resilience and trust while producing measurable returns. Telco-grade data and platforms can help you get there, provided they are used responsibly.
So ask your teams the following:
- Where can a behavioural signal lift approvals without lifting losses
- Where would a fraud risk flag save you the most pain at the least cost
- What would it take to integrate these signals safely into your existing flows
- Which outcomes would you measure weekly to prove the case
Answer those, and AI stops being a talking point and becomes a lever.
As the industry gathers in Pretoria on 24 and 25 November, the conversation will be noisy. Keep your focus on outcomes. Inclusion you can quantify. Fraud you can prevent. Platforms that keep you compliant. Partners who help you orchestrate, not lock you in. That is how AI pays its way in African financial services.
FAQ's
Vodacom Business provides the digital infrastructure that fintechs need to scale – from secure cloud hosting and data connectivity to compliance-ready platforms. Their FinTech Cloud and sovereign AI capabilities allow startups and enterprises to innovate while meeting local data and regulatory requirements.
XLink acts as a payments orchestration layer, connecting merchants, banks, and payment service providers across Africa. It simplifies the complexity of multiple payment methods and ensures secure, compliant transactions locally and cross-border.
Compliance-by-design means embedding regulatory requirements directly into system architecture and product design – not adding them later. This makes it easier for fintechs to adapt to new regulations without disrupting customer experience or product delivery.
A sovereign AI cloud ensures that all customer and transaction data stays within the country of operation, giving organisations full control over data governance. This is crucial for financial institutions and fintechs operating under strict data residency and privacy laws.
Through try-and-buy models, scalable cloud offerings, and orchestration-as-a-service, smaller fintechs can access enterprise-grade infrastructure and AI tools without upfront capital investment. This levels the playing field, allowing startups to compete with larger incumbents in speed, compliance, and reliability.