165,000++183% YoY
annual messages
600+
financial institutions covered
13,800+144% YoY
monthly chats

The National Bank of Rwanda (BNR) is responsible for protecting financial consumers and maintaining trust across Rwanda’s financial system. As digital financial services expanded nationwide, BNR recognised the need for a modern, inclusive mechanism that would allow citizens to report complaints, access authoritative financial information, and track redress without relying on physical visits or fragmented, manual workflows.

To meet this need, BNR partnered with Proto to deploy Intumwa, an AI-powered consumer protection agent. Initially focused on digitising complaints handling, Intumwa has since evolved into a broader conversational utility supporting both complaints management and information delivery across multiple digital channels.

Gérard Nsabimana, Director of Market Conduct Supervision, BNR said: “The National Bank of Rwanda is pleased to deploy the AI-powered grievance redressal system from Proto. The Intumwa AI agent will deliver consumer protection to Rwandans in their local languages and preferred messaging channels.”

A single entry point for the entire financial sector

Before Intumwa, citizens had no standardised digital channel to file a complaint against a bank, insurer, SACCO, or mobile money operator. Routing a case to the right institution required manual triage, and BNR had limited visibility into volumes, trends, or resolution outcomes.

Intumwa replaces this with a structured intake flow on webchat and WhatsApp. The AI agent guides each consumer through a short conversation to capture:

  • The financial institution involved – verified against BNR’s registry of licensed FSPs
  • A description of the issue
  • Contact details for follow-up

Once submitted, the complaint is routed automatically to the correct institution’s queue. The consumer receives a reference number and can check their case status at any time through the same chat interface – without needing to call BNR or visit in person.

Every complaint is logged, timestamped, and visible in BNR’s supervisory dashboard.

Webchat carries the larger share of traffic and keeps growing quickly – volume is already 29% ahead of all of last year’s total within eight months – while WhatsApp, the newer of the two channels, is scaling fast behind it. Combined interactions across both channels have grown 66% since WhatsApp joined the mix, and on its current trajectory the platform is on pace for over 165,000 messages this year, nearly triple last year’s total. Chat volume specifically is up 144% year on year and 250% since the start of last year, reaching 7,300 unique users in the latest month, up from 2,400.

29%
webchat growth vs last year
66%
chat increase after WhatsApp enabled
250%
chat growth since last year

Policy-enforced resolution timelines

Each complaint follows a 15-business-day resolution window, built directly into the platform and enforced automatically across all 600+ institutions. FSP compliance with that standard has been climbing fast – up 17.0 percentage points in a single month to 47.8%, as more institutions adapt to the workflow.

Average resolution time has followed the same trajectory. It briefly rose to 44 days during an earlier platform transition, as institutions adjusted to a new way of working, and has since fallen to 18 days as the sector settles in.

If a deadline is missed, the consumer can escalate directly through Intumwa and BNR can intervene via its internal case management tools. Every response is logged and every breach surfaced automatically – creating a permanent, auditable record of how each institution handles consumer redress.

Kinyarwanda support built for Rwanda’s context

Intumwa operates in Kinyarwanda, English, and French. Kinyarwanda support runs on Proto’s self-trained local language model – built specifically for Rwanda’s linguistic, financial, and regulatory context, not a translated or off-the-shelf system. This delivers higher intent-detection accuracy for the colloquial and mixed-language inputs that characterise how citizens actually describe financial issues, improving:

  • Complaint intake quality
  • Correct routing and categorisation
  • Adoption in rural and underserved communities

Automated categorisation and supervisory intelligence

Intumwa automatically classifies incoming cases by issue type – fraud, loan disputes, failed transfers, and others – at intake, without manual review. This allows BNR to:

  • Identify emerging patterns across institutions
  • Compare issue trends market-wide
  • Prioritise supervisory attention based on evidence

A recent validation update closed a loophole that had let general enquiries get logged as complaints – tickets dropped 84% in a single month as a result, giving BNR a cleaner, more accurate baseline to act on.

47.8%+17.0pp MoM
FSP compliance rate
18 days-59%
average resolution time
84%
ticket-validation drop

The platform also handles general inquiries – consumer rights, financial regulations, NBR services – dynamically switching between information and complaint flows based on user intent. This capability is being extended into BNR’s mobile application, where Intumwa will guide users through digital forms and provide regulatory updates.

Escalation from FSP AI agents to the regulator

BNR is enabling AI agent switching between financial service providers and Intumwa. When a user at an FSP’s own digital channel indicates dissatisfaction or needs to file a formal complaint, the conversation transfers directly into Intumwa – with context preserved – and enters the official regulatory workflow immediately.

A recognised solution, extended into recovery

Proto and BNR have received two international awards recognising Intumwa’s impact.

Noella Kajeneri, Proto’s Managing Director Africa, accepting an award at Accion’s Trust by Design Fintech Challenge

Proto was named one of three global finalists – selected from more than 200 applicants – in Accion’s Trust by Design Fintech Challenge, recognised for excelling across all four of the challenge’s criteria. The Selection Committee singled out Intumwa’s deployment in Rwanda for pairing resolution with visibility: consumers receive a tracking ID and status updates, while regulators receive analytics that support supervisory decision-making. Since launch, the share of complaints filed by women in Rwanda has doubled, from under 10% to 20%.

A consortium of BNR, Proto, Financial Network Analytics (FNA), GLEIF, and the University of Manchester built on that foundation, winning the AI-driven fraud and scams challenge at the 2026 C:>DIR Global ‘Agentic Regulator’ Hackathon in Cambridge. The winning prototype extends Intumwa from grievance redress into fraud recovery: a consumer’s scam report now connects directly to fund tracing, with every output treated as a recommendation until an authorised human approves an irreversible action.

David Butera, Director MIS, National Bank of Rwanda said: “AI can help us move much faster in collecting information, tracing transactions and preparing cases, while decisions such as freezing funds remain with the authorised institutions.”

About FNA

FNA is a leading analytics and simulation firm specialising in financial network mapping, stress testing, and fraud detection. Trusted by central banks, regulators, and financial market infrastructures worldwide, FNA’s platform helps institutions visualise and analyse complex transaction networks in real time. Its flagship solutions – such as the National Fraud Portal and Money Trails – enable authorities to trace illicit fund flows, detect high-risk accounts, and respond to scams with unprecedented speed and precision.

About the National Bank of Rwanda

The National Bank of Rwanda (BNR) is Rwanda’s central bank, responsible for monetary policy, financial sector supervision and regulation, and for advancing financial inclusion across the country’s banking, insurance, pension and payments sectors.

About Proto

Proto deploys inclusive AI workflows in emerging markets. The company is trusted by governments and enterprises to automate workflows for anti-scam centres, patient experience, and other mission-critical usecases. Proto's clients include central banks, remittance services, and hospitals protected with the company's SOC2, ISO27001, GDPR, and HIPAA compliance. Proto's text and voice AI datasets power high performance for local languages beyond the limits of large language models. Headquartered in Canada, Proto operates from regional offices in the Philippines and Rwanda.

15‑minute live demo

Ready to connect with our team?