AI Agents Playbook: Reimagining Financial Services in Africa 

Download Free

Introduction

The Day Insurance Stopped Asking for Photocopies.

iPF Softwares partnered with NMB Bank and Metro Assurance to design an AI-powered WhatsApp solution that transformed life assurance streamlining workflows, boosting customer engagement, and driving enterprise-level operational efficiency in Tanzania’s financial sector.

In the rapidly evolving landscape of banking and finance, staying ahead of technological advancements is crucial for maintaining a competitive edge.

This case study explores how we partnered with a leading bank in Tanzania to develop an AI agent solution that revolutionized their customer service operations, enhanced user experience, and optimized internal processes.

PARTNERSHIP GOAL

→ The goal was to build an AI agent seamlessly integrated with the bank’s core systems delivering 24/7 personalised support, automating routine tasks, and freeing human agents to focus on complex cases, thereby improving customer satisfaction and enterprise efficiency.


About Our Partners

NMB Bank PLC

NMB Bank is one of Tanzania’s largest and most innovative financial institutions, known for its commitment to leveraging technology to enhance customer experiences and drive financial inclusion across the country.

Metro Life Assurance

Metro Life Assurance is a leading insurance provider in Tanzania, offering a wide range of life assurance products and services designed to meet the diverse needs of individuals and businesses.


The challenge

Digitising Life Assurance, One Chat at a Time.

When Metro Assurance and NMB Bank approached iPF Softwares, they needed more than a chatbot—they needed a smarter, conversational layer that could navigate the complexities of group life assurance. This included handling:

  • High-friction, paper-based onboarding processes
  • Poor communication leading to customer drop-offs
  • Disjointed internal workflows and underwriting delays

This case study explores how we partnered with a leading bank in Tanzania to develop an AI agent solution that revolutionized their customer service operations, enhanced user experience, and optimized internal processes.

When we were first challenged to work on the Life Assurance Chatbot, we knew it wasn’t just about automation. It was about building trust, accessibility, and reimagining how insurance could work for everyday Tanzanians.

grayson julius from ipf softwares

Across many African markets—including Tanzania, Zambia, Botswana, and Uganda—life assurance remains largely offline. This solution aimed to reverse that trend, starting with Tanzania.


Setting the Stage

Channel, Approach & People

In a market where WhatsApp is as common as airtime itself, the question wasn’t whether to build a WhatsApp-based solution—it was why we hadn’t done it sooner.

WhatsApp is the most used messaging platform in Africa, with over 300 million active users across the continent (Statista, 2023).

In Tanzania, more than 90% of smartphone users rely on WhatsApp as their primary digital communication tool (TCRA, 2022). The opportunity wasn’t just about digitising insurance—it was about doing so within a familiar channel that customers already trusted.

We envisioned a conversational agent that could:

  • Engage customers directly on WhatsApp
  • Guide group chairpersons through onboarding
  • Capture and validate KYC documents using OCR
  • Support customer queries and underwriting workflows seamlessly
  • Claim submissions and status tracking

All within a simple, conversational WhatsApp interface.


Why WhatsApp? A Strategic Channel

WhatsApp isn’t just convenient—it’s the dominant messaging platform across Tanzania. This made it the ideal digital front door for insurance services.

  • Frictionless onboarding: No new apps or signups
  • Multimedia interaction: Share IDs, photos, and voice notes
  • User familiarity: Already part of customers’ daily habits

Our Approach: Co-Creation and Iteration

We started with people—not technology. Our process involved:

  • Empathy-led Discovery :

    Engaged bancassurance teams and call centre agents to map real frustrations and pain points. Design was informed by language preferences, device limitations, and literacy barriers.

  • Rapid Prototyping :

    Built and tested early WhatsApp flows. We refined OCR handling for IDs and adapted tone and language for clarity—leveraging Swanglish where appropriate.

  • Smart Integration :

    Created a microservices architecture that securely connected the agent to underwriting systems, support tools, and payment APIs.

  • Live Learning:

    Post-launch analytics guided real-time improvements. The AI agent continues to learn from conversations, enhancing accuracy and relevance over time.


The Execution Team: Built by a Cross-Functional Powerhouse

The project was delivered by a dedicated, cross-functional team at iPF Softwares, including AI engineers, backend developers, UX experts, and project managers. Working closely with Metro Life and NMB Bank, the team co-designed and implemented a chatbot that reflected real user needs while aligning with local regulations and insurer workflows.


Deep Dive

Engineering the solution

OCR Document Processing

Our open-source OCR engine was customised to extract data from Tanzanian national IDs, policy documents, and claim forms. OCR performance was tuned to low-resolution image uploads—often sent via WhatsApp camera. Key techniques included:

  • Region-based text extraction (RBTE) for capturing name, NIN, DOB fields
  • Confidence scoring and fallbacks for error-prone handwriting or lighting conditions
  • Language model reinforcement to improve field matching and reduce false positives

Seamless Integration Layer

Integrating with both NMB Bank’s and Metro’s systems meant juggling:

  • Core BancAssurance platforms (policy, products & claim systems)
  • CRM tools for existing customer context
  • Payment integration with bank's payment gateway for premium collections

We built a robust middleware layer using FastAPI and Nest JS, exposing RESTful APIs that bridged WhatsApp interactions with backend systems. Key features included:

  • Reusable API adapters for fast service mapping
  • Audit-logging and retry queues for reliability
  • Custom failover handling for system unavailability

Conversation Orchestration

The AI Agent wasn’t just scripted—it could reason across states. We implemented:

  • Context-persistent threads across sessions (e.g., onboarding vs claims)
  • Smart interruptions: if a user changed the topic, the bot could switch flows
  • Human escalation logic based on sentiment, confidence, and intent overlap

The orchestration layer was designed in FastAPI with fallbacks to FAQ responses when the LLM couldn’t confidently route.


The Solution

AI That Speaks and Serves

An AI Agent Tailored for Group Life Assurance

Our solution was a custom-built AI agent integrated directly into WhatsApp, designed to handle the entire group life assurance journey—from onboarding to claims. Key features included:

  • Conversational Onboarding: Guided group chairpersons through policy selection, member addition, and document submission via an intuitive chat interface.
  • OCR Document Processing: Automated extraction and validation of KYC documents using a custom-trained OCR engine, reducing manual data entry errors.
  • Real-time Query Handling: Leveraged NLP to understand and respond to customer inquiries about policy details, coverage, and claim status.
  • Seamless Integration: Connected with NMB Bank’s and Metro Assurance’s backend systems for real-time policy management, underwriting workflows, and payment processing.
  • Claim Submission & Tracking: Enabled customers to submit claims directly through WhatsApp, with real-time status updates and document uploads.

This AI agent not only streamlined operations but also provided a user-friendly experience that resonated with the Tanzanian market.


Technology Stack

The solution was built using a combination of:

  • WhatsApp Business API for seamless messaging
  • Custom OCR engine built on Tesseract and OpenCV
  • NLP models fine-tuned for insurance-specific intents and entities
  • RESTful APIs for integration with banking and insurance systems
  • OnPremise infrastructure for data security and compliance

This robust tech stack ensured the AI agent was not only powerful but also scalable to handle growing user demands.


Final Thoughts

This project wasn’t just about building a chatbot. It was about transforming how insurance is delivered—replacing bureaucracy with conversation.

By meeting customers where they are—on WhatsApp—we helped NMB Bank and Metro Assurance lead the way in digital-first, human-centred insurance across Africa. The result is a scalable, smart, and inclusive solution that redefines what’s possible in modern bancassurance.

background-image

CONTACT US

Need a custom AI solution?

Reach out to us for a personalized consultation to explore how our AI solutions can be tailored to meet your unique business needs.

Whether you're looking to enhance customer experiences, streamline operations, or drive innovation, our team is here to help you navigate the possibilities of AI.

By submitting this form, you agree to allow www.ipfsoftwares.com to store and process the personal information submitted above to provide you the content requested.

Ready to transform your business?

Let's make something impactful together

Let's Talk