Why A Chatbot That Speaks Karanga Matters More Than A Chatbot That Speaks English

By Ross Moyo

Shona is spoken by 14 million people, yet for years AI treated it as if it did not exist. Shona 2.0 is not just about faster customer service. It is about who gets to be heard – in their own tongue, in their own dialect – by the machines that now run daily life.

For years, talking to a machine in Shona felt like talking to a robot. Now Econet says its chatbot understands 8 out of 10 queries, handles Shona-English mixing, and will soon tell a Zezuru speaker from a Karanga speaker – and answer back in their own dialect.

Econet Ai – a wholly owned subsidiary of Econet Wireless Zimbabwe – has launched Shona 2.0, the latest version of its Shona Artificial Intelligence (AI) model.

AI models power Yamurai, Econet’s multilingual customer support chatbot, with the intelligence that enables it to increasingly handle most of the company’s customer inquiries.

Econet said the model now has the capability to comprehend and answer 80% of customer queries, putting it on par with international languages, and marking the first time an African AI language model has reached this level of comprehension.

That last claim matters. If it holds up to independent testing, it is not just an Econet upgrade. It is a continental milestone.

For decades, African languages have been classified in AI research as low-resource – meaning there is not enough clean, labelled text, speech and conversation data to train models to the same standard as English, French or Mandarin. African languages remain underrepresented in natural language processing, with most corpora limited to formal registers that fail to capture the vibrancy of everyday communication.

Econet is now saying: we have closed that gap, at least for customer service in Shona.

To understand why Econet Ai’s announcement matters, you have to understand the history of Shona in Zimbabwe – and the history of AI in Africa.

Shona, or chiShona, is the most widely first spoken Bantu language, native to the Shona people of Zimbabwe.

According to Ethnologue, Shona, comprising the Karanga, Zezuru, and Korekore dialects, is spoken by about 10.8 million people. The Manyika and Ndau dialects of Shona are listed separately by Ethnologue, and are spoken by 1,025,000 and 2,380,000 people, respectively. The total figure of Shona speakers is then about 14.2 million people.

Other counts put it at L1: 9.6 million (2010-2023), L2: 4.5 million (2023), Total: 14 million.

By any count, Shona is one of the largest Bantu languages in the world. Zulu is the second most widely spoken Bantu language with 10.3 million speakers according to Ethnologue.

And yet, until very recently, if you asked ChatGPT, Gemini, Claude or any global model to help you in fluent, idiomatic Shona – with slang, with proverbs, with dialect – it would stumble, translate literally, or switch back to English.

Why? Because AI learns from text. And African languages have very little text online, relative to their speaker numbers.

Africa has over 2,000 languages. Building AI that works in this context requires multilingual architecture from inception, not as an afterthought.

When Vambo AI, a Johannesburg-based AI model builder, set out to build MORENA, a 1.5-billion-parameter model covering 12 African languages, it had access to a supercomputer through a UNDP-backed programme. It had the GPUs, specialised chips that power AI computing, but struggled to find enough real African-language text.

Finding that much real data was close to impossible, so a lot of it was synthetic, Isheanesu Misi, Vambo’s co-founder and chief technology officer, told TechCabal.

The shortage of African-language text data threatens to limit the development of AI tools that reflect the continent’s linguistic diversity. Without enough high-quality data, developers risk building AI products that perform poorly in local languages, leaving millions of potential users underserved.

Misi said languages such as isiNdebele and Nigerian Pidgin can have next to no resources available for AI development.

That is the gap Econet is trying to close – not by scraping the internet, but by listening to its own customers.

Econet has 12.9 million active subscribers. According to the First Quarter 2026 Sector Performance Report published by POTRAZ, Econet added over 552,000 active subscribers, expanding its customer base from 12,374,206 in Q4 2025 to 12,926,399 in Q1 2026. That elevated Econet’s market share from 73.75% to 74.80%.

Every one of those customers generates conversations: WhatsApp messages, calls, complaints, questions. If those conversations – anonymised, consented, cleaned – become training data, Econet has something no global lab has: millions of real, code-mixed, dialect-rich Shona conversations about real problems.

That is what Kapur means when he says: AI models are trained just like children; they learn and continually improve from listening to our customers.

What Shona 2.0 actually does

According to Econet, Shona 2.0 can now understand customers who mix Shona and English when speaking, a common practice that often confuses AI Models.

Anyone who has lived in Harare, Bulawayo, Mutare or Masvingo knows that practice. It is not bad Shona or bad English. It is how Zimbabweans actually speak:

Ndoda ku-checker data balance yangu, but bundle rangu rapera?
Ndaka-recharger airtime but haisi ku-reflecter, mubatsirei?
PUK number yangu ndaikanganwa, mungan-di-send-er here?

That code-mixing – Shona-English slang, verb borrowings, English nouns with Shona prefixes – is exactly what breaks most global models. They were trained on formal corpora, struggling with slang and code-mixed inputs.

A 2025 study of Shona-English slang found the same problem and built a dataset from anonymised social media conversations, annotated for intent, sentiment, dialogue acts, code-mixing, and tone, to prove it could be solved. That work fine-tuned a multilingual DistilBERT classifier for intent recognition, achieving 96.4% accuracy and 96.3% F1-score.

Econet’s claim is different in scale. It is not a research demo. It is a production system answering real customers, on WhatsApp, 24 hours a day, about real money – airtime, bundles, PUKs, SIM swaps, roaming.

Yamurai is the WhatsApp Chatbot Virtual Assistant from Econet Wireless, reachable by sending messages to the WhatsApp number +263 781 222 000.

This multi-lingual assistant can communicate in Shona, Ndebele, English, and other recognised languages. It offers guided interactions through predefined buttons, making it easy for customers to access automated services 24/7. It helps with tasks such as SIM replacement, international prepaid roaming, checking data bundle balances, retrieving PIN and PUK codes, addressing over-scratched cards, managing out-of-bundle browsing, reporting network issues, and transferring airtime.

If Shona 2.0 can truly handle 80% of those queries in Shona, in Shona-English mix, without handing over to a human, that is a transformation in service delivery for a company that dominates Zimbabwean telecoms.

The dialect breakthrough

Econet Ai has also begun to train the model in six different Shona dialects. It will soon be able to identify the dialect a customer is using and automatically respond in that dialect.

This is the most culturally significant part of the announcement – and the most technically difficult.

The Shona language is spoken in five dialectal clusters, which are ChiKaranga, ChiManyika, ChiNdau, ChiZezuru, and ChiKorekore. ChiKaranga is spoken in areas surrounding Masvingo town in the south, ChiManyika in areas around Mutare in the east, ChiNdau in Chipinge in the south east, ChiKorekore in Mutoko in the north east, and Chizezuru in northcentral Zimbabwe.

It should be noted that Shona dialects are mutually intelligible and are, thus, dialects of the same language. Indeed, 80 to 90 per cent of the vocabulary is common to the whole area of Shona speaking communities.

But that 10% to 20% difference is where dignity lives.

For decades, school curricula, textbooks, examinations, and teacher training privileged Standard Shona (Zezuru/Karanga core). Students from Manyika, Ndau, or Korekore backgrounds who used their home lexical items, phonological features, or morphological patterns were marked wrong.

A diglossic situation emerged. Zezuru occupied the highest prestige position, Karanga an intermediate but still elevated status, and Manyika, Ndau, and Korekore lower status. Speakers of the latter were often ridiculed for incorrect, rural, or funny-sounding speech.

Econet is now promising to reverse that logic in a chatbot: instead of forcing the customer to speak Standard Shona to be understood, the machine will learn to hear the customer as they are – and answer back in that dialect.

Econet has not yet published the list of the six dialects it is training on. The five canonical clusters are Zezuru, Karanga, Manyika, Ndau and Korekore. The sixth may be a sub-dialect – such as Budya, or a distinct urban Harare variety – or Kalanga, which some researchers treat as part of the Shona cluster. This newspaper has asked Econet Ai to confirm the six.

What matters is the principle: dialect is not an error to be corrected. It is data to be learned.

Trained just like children

AI models are trained just like children; they learn and continually improve from listening to our customers. So, you find that in the beginning Yamurai in Shona was like speaking to a robot, but today it is getting much more like a person. It is getting better all the time, said Econet Ai CEO Mr Navdeep Kapur.

That metaphor is more accurate than it sounds.

Large language models improve not by being programmed with grammar rules, but by being exposed to millions of real conversations, corrected, and retrained. Econet says there is a full-time engineering team working on continually training and improving the Yamurai language models.

That team is part of a much larger build-out. Econet AI recruited more than 100 engineers and specialists. Its partnership with Cassava AI provides access to NVIDIA accelerated computing infrastructure, enabling practical AI solutions across telecommunications, finance, agriculture, insurance, mining, healthcare, retail, logistics, education and government.

In other words, Yamurai is not a side project. It is the front door of a new AI division.

The colonial history of Standard Shona

Econet’s decision to train in six dialects is not just technical. It is historical redress.

In 1929-1931 the Southern Rhodesian government and the Carnegie Corporation commissioned South African linguist Clement Martyn Doke to resolve competing missionary orthographies and produce a unified writing system. His Report on the Unification of the Shona Dialects (1931) classified the main clusters as Zezuru, Karanga, Manyika, Ndau, and Korekore.

Doke recommended a single orthography and literary standard based primarily on Zezuru, with substantial input from Karanga, and progressively less from Manyika, Ndau, and Korekore.

The designation of Zezuru as the standard Shona and also the capital city of the country being in the Zezuru speaking region made Zezuru the superior language.

The result, as one recent analysis puts it: Colonial language planning under Doke did not invent difference; it ranked existing variation and institutionalised that ranking through schools, literature, and administration.

For a child in Chipinge who says ndakabva ku… in Ndau, or a child in Mutoko who says tedja in Korekore, school taught them that their home language was wrong. The correct form was the Zezuru form in the textbook.

Now, a machine is being taught the opposite: that the home form is valid, that it should be recognised, and that it should be answered in kind.

If Econet delivers on that promise – if a Manyika customer gets a Manyika answer, a Korekore customer gets a Korekore answer – it will be the first mass-market technology in Zimbabwe to treat dialects as equals, not as deviations.

Ndebele is next

Ndebele 1.0 was launched last month, and work is at an advanced stage to get Ndebele 2.0 out before the end of the month.

That timeline is aggressive – two major language upgrades in less than 60 days – and it signals Econet’s sequencing: Shona first, because it is the largest language; Ndebele second, because it is the second largest; then, according to earlier Econet statements, Tonga, Kalanga and other languages spoken in Zimbabwe.

Econet is currently the only mobile operator in Africa with a multilingual AI helpline offering support in local languages. Its Yamurai service answers questions in English, Shona and now Ndebele, with Tonga, Kalanga and other languages spoken in Zimbabwe set to be added soon – a critical inclusion strategy for rural and elderly users.

That claim – only operator in Africa – has not been independently verified by this newspaper, but no other African operator has publicly demonstrated a production chatbot handling two indigenous languages at this scale.

The company said Yamurai had drastically improved Econet’s ability to respond to customer queries and significantly reduced customer waiting times on the helpline, ensuring that all queries are addressed.

The scale of that improvement has been reported before: Econet AI’s customer service platform, Yamurai, has cut response times from between eight and ten minutes to just three seconds, effectively giving each customer a personalised digital assistant.

Earlier reporting also noted that Econet’s Shona-speaking chatbot, Yamurai, can already handle more than 70 per cent of customer queries, alongside about 600 human contact-centre staff.

Econet’s new figure – 80% – therefore represents a jump from 70%+ to 80% in comprehension. In a call centre that serves millions, a 10-percentage-point jump means hundreds of thousands fewer people waiting on hold.

Real light at the end of the tunnel

Econet is continuously expanding Yamurai’s capabilities as it seeks to improve the quality of its service delivery. Through Econet Ai, the company is tackling problems many operators in Africa have struggled with for years without any real solution.

Now there is real light at the end of the tunnel, as AI opens solutions at all levels of telco operations. It takes time to get them all deployed but we are confident that we are only months away from having the best service in Africa, said Mr Kapur.

Econet said the development of multilingual and agentic chatbots forms part of its broader strategy to use AI to improve customer service, increase accessibility and provide faster support across its platforms.

Agentic is the key word there. It means a chatbot that does not just answer – Your balance is X – but acts – I have recharged you, I have blocked the lost SIM, I have escalated the network fault to the team in your area.

That is the next frontier for Yamurai: from answering to doing.

The 16-language test

Zimbabwe’s Constitution recognises 16 official languages, reflecting the country’s rich cultural heritage. These include Shona, Ndebele, Tonga, Kalanga, Nambya, Venda, Sotho, Shangani, Chewa, Chibarwe, Xhosa, English, Sign Language, Tswana, Koisan and Ndau.

Since the adoption of its 2013 Constitution, Zimbabwe has 16 official languages, namely Chewa, Chibarwe, English, Kalanga, Koisan, Nambya, Ndau, Ndebele, Shangani, Shona, sign language, Sotho, Tonga, Tswana, Venda, Xhosa. The country’s main languages are Shona, spoken by over 70% of the population, and Ndebele, spoken by roughly 20%.

Each language represents more than a means of communication. It carries historical knowledge, cultural values, traditional practices and unique ways of interpreting the world.

This presents an opportunity to establish a coordinated national indigenous language digitisation programme aligned with Zimbabwe’s National Artificial Intelligence Strategy 2026-2030.

Importantly, priority should not be limited to Shona and Ndebele. Languages with relatively limited digital representation, including Tonga, Kalanga, Nambya and others, deserve deliberate attention to prevent technological development from reinforcing existing inequalities.

Imagine a grandmother in Matabeleland accessing healthcare information in Ndebele, or a student in Binga learning science through Tonga without struggling to translate complex concepts from English.

That is the test for Shona 2.0: will it be a one-language wonder, or the template for all 16?

Econet says the latter. It says Tonga, Kalanga and other languages are set to be added soon. It has already launched Ndebele 1.0 and is working to get Ndebele 2.0 out before the end of the month. That suggests a pipeline, not a one-off.

But the pipeline will need data, engineers and money for languages with far fewer speakers and far less digital text than Shona. For Tonga in Binga, for Kalanga in Plumtree, for Nambya in Hwange, for Ndau in Chipinge – the data problem is even harder than for Shona.

If Econet can solve it for Shona dialects, it will have a method to solve it for smaller languages. If it cannot, the risk is that AI will deepen the same inequality the 2013 Constitution tried to fix: Shona and Ndebele get AI, everyone else gets left behind.

Code-mixing is not a bug. It is the language.

The most under-appreciated part of Econet’s announcement is code-mixing.

Global AI benchmarks often test models on clean Shona – textbook Shona, Bible Shona, news Shona. Real Zimbabweans do not speak clean Shona. They speak Shona-English, Ndebele-English, Shona-Ndebele-English, with slang, with WhatsApp abbreviations, with voice notes.

Research on African languages on social media found that when these languages are used, they are often not in a conversational context. Instead, their usage is limited, primarily appearing in questions or for translation purposes, with a heavy prevalence of English.

Language detection models, including the specialized AfroLID and a general LLM, performed with near-perfect accuracy on the clean news data but struggled with the code-switched Reddit posts.

Econet’s model is trained on the opposite data: messy, real, code-switched customer conversations. That is why it can now, according to Econet, understand customers who mix Shona and English.

That is also why the 80% figure matters. It is not 80% on a clean benchmark. It is 80% on real customer queries, in real language, about real problems.

This newspaper has not independently tested Shona 2.0. Econet has not published a benchmark, a test set, or a methodology for the 80% figure. For the claim to be fully credible, Econet Ai should publish:

1. How comprehension is defined – is it intent recognition, or full resolution without human help?
2. What the test set was – how many queries, in which dialects, in which mix?
3. What the error rate is for each dialect – does Karanga get 85% and Ndau get 65%?
4. How human review works – what happens to the 20% it does not understand?

Those are not criticisms. They are the questions any serious newspaper should ask when a company claims a continental first.

From call centre to AI company

To understand Shona 2.0, you have to understand what Econet has become in 2026. Econet is no longer just a mobile network. With 100+ AI engineers, a 100-model marketplace, and a chatbot that now speaks Shona better than most global models, it is trying to do for AI what EcoCash did for mobile money.

Econet Wireless Zimbabwe recorded revenue growth of 21 percent for the half-year ended August 31, 2026, supported by sustained demand for connectivity, insurance and financial services.

Performance across its four strategic pillars – Mobile Network Operations, Financial Technology, Econet Infrastructure and Econet AI – strengthened the group’s growth, resilience and integrated digital ecosystem.

Econet chairman Dr James Myers said artificial intelligence had become an increasingly important part of the group’s business strategy, moving from experimentation into core operations and product development.

Data traffic increased by 60 percent and voice traffic by 21 percent.

The group accelerated artificial intelligence adoption across customer service, network optimisation, operational efficiency and product development, while pursuing new revenue opportunities.

In other words, AI is no longer a pilot. It is a pillar, alongside the network itself.

The four pillars in practice are: 1. Mobile Network Operations: Econet ended Q4 2025 with 340 5G base stations and 7,143 total active base stations, exceeding NetOne’s 4,838 and Telecel’s 1,123 combined network footprint. Econet holds 73.75% of active mobile subscribers, 88.32% of mobile voice traffic, and about 81.2% of mobile internet and data traffic. By Q1 2026, Econet’s mobile data usage rose by 20.14% to reach 156.4 Petabytes, bringing its mobile data market share to 87.21%. 2. Financial Technology: EcoCash remains the backbone of Zimbabwe’s digital payments. Of 234.7 million electronic payments recorded in the first quarter of this year, around 208.8 million went through mobile wallets, most of them on EcoCash. 3. Econet Infrastructure: The company is building the towers, fibre and data centres that other operators and enterprises will rent. 4. Econet AI: This is the new pillar. And Yamurai is its most visible product.

The EcoCash playbook

Econet is deliberately replicating the playbook it used to turn EcoCash from a novel product in 2011 into Zimbabwe’s dominant mobile money platform now used by millions.

EcoCash worked because it solved a real problem – cash shortages, bank queues, sending money to rural areas – in a language people understood, on phones they already had, at a price they could afford.

Econet Ai is trying to do the same for AI.

We want everyone who has a smartphone to start using AI, and that is nearly 70% of our base. Unfortunately, you can’t access AI without a proper smartphone, said Econet Ai CEO Mr Navdeep Kapur.

Zimbabwe has over 15 million mobile subscriptions and smartphone penetration is estimated at between 65-70%, one of the highest in Southern Africa, providing a fertile ground for the campaign.

Econet says its ambition is to make Zimbabwe the most AI-savvy country in Africa, matching countries like China, Singapore and the US where up to 70% of citizens use AI on a daily basis.

Econet Group Chief Executive Officer Dr Douglas Mboweni said Zimbabwe already had what the company believes to be the highest level of daily AI interaction in Africa, supported by services such as Yamurai, EcoChat and EcoFarmer.

Zimbabweans are using AI every day through Econet products like Yamurai, EcoChat, and EcoFarmer.

Yamurai, EcoChat – Econet’s AI customer care assistant – and EcoFarmer, which provides agronomic advice to farmers, have already introduced millions of Zimbabweans to AI without them necessarily knowing it.

That is the EcoCash insight: people do not adopt AI. They adopt a solution that helps them check a bundle, send money, get farming advice, find a Bible verse, help a child with homework.

CAIMEX: 100 models, one platform

Central to the campaign is CAIMEX, Econet’s flagship AI marketplace platform described as the first of its kind in Africa. CAIMEX gives users access to over 100 curated global AI models – including large language models, image generators, coding assistants, translation and productivity tools – in one place.

The platform operates through a token-based payment system, potentially creating a new revenue stream for Econet while giving consumers access to multiple AI models without subscribing individually to each provider.

Users access the models by purchasing AI tokens which power their interactions with the tools.

CAIMEX also includes EcoAI Tokens, providing access to ChatGPT, Gemini, Claude and Mistral through one bundled subscription. Google Gemini was introduced free of charge to eligible customers, expanding access to productivity tools.

The logic is simple: a Zimbabwean student should not need four subscriptions – ChatGPT + Gemini + Claude + Mistral – to do homework. A small business should not need to pay in dollars for each model. One token bundle, one platform, in Shona, in Ndebele, in English.

That is also why Yamurai matters for CAIMEX. Yamurai is the on-ramp. A customer who learns to talk to Yamurai in Shona on WhatsApp is a customer who is ready to try EcoChat, EcoFarmer, CAIMEX.

The Cassava connection

Econet Ai works closely with Cassava AI, the artificial intelligence business of Cassava Technologies, to bring advanced AI solutions to Zimbabwe. The collaboration connects Econet Ai’s local expertise with Cassava AI’s global technology partnerships and enterprise AI capabilities, supporting the development and delivery of solutions tailored to the Zimbabwean market.

Econet AI’s partnership with Cassava AI provides access to NVIDIA accelerated computing infrastructure.

Cassava had invested in Africa’s first AI Compute Centre to serve the region. Econet AI will roll out Cassava AiCloud, a platform that allows Zimbabwean enterprises to run data-intensive workloads using high-performance computing powered by NVIDIA GPUs. The infrastructure is hosted at the Cassava AI Factory in South Africa, currently the only facility of its kind on the continent, enabling firms to deploy AI tools without building their own systems.

Navdeep Kapur, Econet AI CEO, said the unit will focus on scalable AI systems tailored to African markets. Our vision is to build AI that is not only powerful, but relevant to African contexts, Kapur said.

Kapur said their focus is not to following global hype around AI but to deploy practical solutions that reduce costs, improve efficiency and grow businesses while delivering tangible benefits to ordinary citizens.

Our promise is that we do not want Zimbabwe to be a consumer of AI. We want Zimbabwe to be a creator of AI.

That promise is being tested in the most practical way: can a Zimbabwean team, using African compute, train a model in an African language, to a standard that matches global models?

Shona 2.0 is Econet’s answer: yes, for customer service, for 80% of queries, in Shona.

What best service in Africa would actually mean

Now there is real light at the end of the tunnel, as AI opens solutions at all levels of telco operations. It takes time to get them all deployed but we are confident that we are only months away from having the best service in Africa, said Mr Kapur.

Best service in Africa is a bold claim. What would it look like in practice?

No waiting: If Yamurai handles 80% in Shona, and Ndebele 2.0 does the same in Ndebele, then the 600 human agents can focus on the 20% that are complex, emotional, or high-value – a lost SIM for an elderly grandmother, a network fault affecting a whole village, a fraud case.

No language barrier: A farmer in Binga can ask in Tonga. A grandmother in Plumtree can ask in Kalanga. A student in Chipinge can ask in Ndau. No one has to translate their problem into English first.

Agentic action: Yamurai does not just tell you your PUK. It retrieves it, sends it, and confirms it. It does not just tell you there is a network fault. It logs it, escalates it, and tells you when it will be fixed.

Network that heals itself: Econet has partnered with Ericsson to modernise its mobile network using AI, 5G Advanced technologies, and next-generation IoT services. The company has been steadily integrating AI and machine learning into its operations as it moves toward the creation of more autonomous networks – systems that can configure, optimise, and repair themselves with minimal human intervention.

That last point is crucial. AI at Econet is not just Yamurai. It is network optimisation, fraud detection, predictive maintenance, energy management for base stations, and customer churn prediction.

If AI can predict that a base station in Masvingo will fail tomorrow, and fix it today, that is better service than any chatbot.

The risks – jobs, privacy, and trust

No honest newspaper can write about AI in a call centre without asking about jobs.

Earlier reporting noted that Yamurai works alongside about 600 human contact-centre staff.

If Yamurai goes from 70% to 80% to 90% resolution, what happens to those 600 jobs?

Econet has not said it will cut staff. Its argument is that Yamurai ensures that all queries are addressed – meaning humans were missing queries before, because they could not keep up with volume. AI handles the volume, humans handle the complexity.

That argument is plausible, but it needs to be tested. Zimbabwe’s banking sector has already seen at least 516 banking jobs lost in 2025 alone, according to the Zimbabwe Banks and Allied Workers’ Union. AI is part of that story globally – Barclays said it would cut about 5,000 jobs, roughly 5 per cent of its 93,000 staff, and that 1,000 had already gone.

Econet should publish: how many contact centre staff it employs now, how many it employed before Yamurai, and how their roles have changed – from answering repetitive queries to handling complex cases, with higher pay and training.

The second risk is privacy. Yamurai learns from listening to our customers. That means customer conversations are training data. Econet must be transparent: are conversations anonymised? Are they used with consent? Are they stored in Zimbabwe or in South Africa at the Cassava AI Factory? Who has access? How long are they kept?

The third risk is trust. If Yamurai gets a query wrong – if it tells a customer their bundle is active when it is not, or gives the wrong PUK, or misinterprets a dialect – who is accountable? Econet says Yamurai ensures all queries are addressed. It should also say: Yamurai will tell you when it is not sure, and hand you to a human, in your language, without you having to repeat yourself.

What to watch next

1. Ndebele 2.0 before end of November: Econet says work is at an advanced stage. If it launches on time, with the same 80% claim, that will prove Shona 2.0 was not a one-off.
2. Dialect list: Which six Shona dialects? Will Ndau and Korekore get equal accuracy to Zezuru?
3. Independent benchmark: Will Econet publish a test set, or allow UZ, NUST, or an independent lab to test Shona 2.0?
4. Tonga, Kalanga, Nambya: When will they launch? Those are the languages that will prove Econet’s commitment to all 16 official languages, not just the two largest.
5. Agentic Yamurai: When will Yamurai move from answering to acting – completing SIM swaps, refunds, network fault reports end-to-end, without a human?

For now, the headline stands: a Zimbabwean company, using Zimbabwean engineers, Zimbabwean data, and African compute, has built a model that speaks Shona like a person, not like a robot.

That is not just a telco upgrade. That is a statement about whose language counts in the age of AI.

For now, the direction is clear: the future of AI in Zimbabwe will not be English-only. It will be Shona, Ndebele, Tonga, Kalanga – and it will be mixed, just like we speak.

Econet Ai was launched earlier this year to deliver advanced AI solutions to the Zimbabwean market.