By:
Nana Appiah Acquaye
For
most of Africa's insurers, the cost of picking up the phone has long outweighed
what a customer is worth. That calculus is changing as agentic voice AI drives
the price of a full customer interaction down to a fraction of a dollar.
During
a recent webinar hosted by sustainability-and-AI firm sustainical and its
Ghanaian partner ShrinQ, a live demonstration showed an AI voice agent handling
a full insurance renewal call in roughly two and a half minutes. The system
greeted a caller renewing a policy on his 2022 Toyota Land Cruiser, corrected a
mis-stated policy number, laid out third-party and comprehensive renewal
options at 650 and 2,400 Ghana cedis respectively, took payment, and issued a
new policy valid through mid-2026 all without a human on the line.
The
interaction, presenters said, cost the insurer between 30 and 50 cents to run.
A different cost structure
According
to figures the sustainical team presented, citing McKinsey and other industry
analysts, an agentic voice call runs 30 to 50 cents, compared with $6 to $12
for a call handled by a live agent. Independent contact-centre research broadly
supports the direction of that gap, even if the exact figures merit scrutiny.
That
shift matters because the traditional cost floor of reaching customers staffed
call centres, physical branches has long kept insurance and banking out of
reach for millions of people across the continent. A voice agent carries no
such floor: it does not staff shifts, and its per-call cost does not rise with
volume.
The
scale of the reach problem is significant. In Nigeria, insurance penetration
sits below 1 percent of GDP, among the lowest rates globally, according to the
country's insurance regulator, NAICOM a market that is largely unserved rather
than merely underserved.
Momentum without full adoption
More
than 40 percent of African institutions have already begun experimenting with
or implementing generative AI, according to McKinsey's May 2025 report Leading,
not lagging: Africa's gen AI opportunity. The same report estimates gen AI
could unlock $2.1 billion to $3.2 billion in value for African insurers
specifically, and names Ghana among the markets where experimentation is
growing.
Voice
applications, however, have lagged behind a gap the webinar's presenters framed
as opportunity rather than deficiency, arguing that regulatory appetite and
infrastructure are largely in place already.
The
broader financial technology backdrop supports that view. Projections cited
during the presentation suggest African fintech revenues could grow fivefold,
from around $10 billion in 2023 to $47 billion by 2028. Mobile phone ownership
stands at 86 percent of adults worldwide, according to the World Bank's Global
Findex 2025, and mobile money is a daily habit in Ghana rather than a novelty.
Why language matters
The
presenters argued that voice technology succeeds where app-based platforms have
struggled because it does not assume literacy, smartphone access, or comfort
with English or French menus.
They
pointed to a Nigerian deployment in which an AI agent serves customers fluently
in Yoruba a language spoken by tens of millions but rarely available at a
service desk. The platform demonstrated during the webinar handles Twi, Hausa,
Swahili, Yoruba, Arabic, English and French.
Renewals
were chosen as the initial use case because they represent a persistent weak
point for Ghanaian insurers.
"Renewals
are the single biggest problem facing insurers in Ghana across life, non-life,
pensions and health, everyone struggles to get customers to come back,"
said Stephen Ameyaw, CEO of ShrinQ Limited, the Accra-based firm partnering on
the rollout. "When an agent calls people back in their own language, the
pilots we've run pushed renewal rates to more than double what they are today.
That isn't a cost-cutting exercise. It's a growth strategy. And done right, a
genuine inclusion strategy."
The sovereignty question
The
rollout raises a separate issue that African policymakers have increasingly
sought to address: where customer voice data among the most sensitive personal
data that exists is processed, and under whose jurisdiction it falls.
Routing
insurance calls through foreign cloud infrastructure and standard commercial
models risks recreating a sovereignty problem while solving an inclusion one,
given that data hosted on hyperscaler platforms can be subject to foreign
statutes such as the US Cloud Act regardless of where the customer lives.
Continental frameworks including the African Union's Malabo Convention and the
AfCFTA Digital Trade Protocol have been developed partly in response to that
risk.
The
architecture presented during the webinar addresses this directly. The voice
layer that greets callers has no access to internal data; the underlying agents
that check policies and process payments sit behind a separate boundary and can
run on local models, on an institution's own premises, meaning no part of the
conversation needs to leave its systems. Presenters said the setup does not
require large hardware investment and can run on a single desktop-class AI
workstation.
"The
voice your customer hears is just the edge. It holds no data of its own,"
said Karsten Samaschke, co-founder and CTO of sustainical, whose background is
in technology and sovereign-cloud infrastructure. "The intelligence sits
behind a boundary you control, and it can run on local models, on your own
premises, end to end, so the conversation never has to leave the country. We
can host it inside a specific country, on-prem, or hybrid. Which simply means
you stay in control. We believe in security, and we believe in
sovereignty."
The
trade-off is not without cost. The 30-to-50-cent figure is largely a marginal
cost and tends to be cheapest on the same hyperscaler clouds the sovereignty
argument warns against; on-premise hardware, model tuning, integration and
staff for escalations represent real upfront investment. What sovereignty buys,
on this account, is control and auditability rather than a lower per-call price
— a trade-off whose value will vary between a continental bank and a mid-sized
Ghanaian insurer.
Adoption prospects in Ghana
Whether
Ghanaian institutions take up the technology depends on both how readily the
market adopts it and what value it delivers once running.
Ghana's
Commissioner of Insurance, Dr. Abiba Zakariah, has pointed publicly to
technology as a way to overcome the industry's high cost of establishing
physical offices region by region, and has said the National Insurance
Commission is committed to products suited to the financial realities of
informal-sector workers. Speaking at the UNDP-backed Inclusive Insurance
Innovation Challenge in 2025, she linked that push directly to expanding
coverage among underserved Ghanaians.
"In
Ghana the real question is always how quickly a new platform gets adopted. The
piloting stage is where it's won or lost," said Franklin Amankora Nkrumah,
ShrinQ's Ghana-based partner on the project. "What makes this different is
that it meets the market where it already is: mobile-first,
mobile-money-native, and served in local languages. Start with renewals, prove
the value, and adoption follows."
A partnership model
The
presenters argued that neither the technology nor the market relationship
succeeds alone that only "global AI
capability plus genuine local partnerships" sustains in emerging markets.
The webinar itself modelled that pairing: a UK- and Germany-based technology
firm working with a Ghanaian partner supplying local relationships, regulatory
fluency and on-the-ground understanding.
The
system is designed to run in hybrid mode, handling routine calls while
escalating to human staff when needed, and the same model is intended to extend
beyond insurance into banking, telecoms, utilities and healthcare.
This
article draws on a sustainical–ShrinQ webinar on agentic voice AI in financial
services, sustainical's whitepaper on digital sovereignty prepared for the
Africa Fintech Summit, and input from ShrinQ's Ghana-based team.
