How Lumi Is Bringing AI to Feature Phones in Africa

Learn how Emmanuel Ogolla built Lumi, an offline-first multilingual AI assistant that works on SMS, and why designing for constraints can create better products. If you care about digital inclusion, African innovation, or building with limited resources, this story is for you.

When you think of AI, you probably think of smartphones, apps, and fast internet. But what about the millions of people still using feature phones, relying on SMS, and living with unreliable connectivity?

That is the gap Emanuel Ogolla set out to close with Lumi, an offline-first AI assistant built for people technology often forgets. In this conversation, he explains how he went from having zero coding experience to building a multilingual AI product that works over SMS in English, Swahili, and Sheng - and why that constraint became the product’s biggest strength.

What follows is more than a founder story. It is a practical lesson in building for real users, especially when your audience lives outside the usual startup assumptions.

Why Building for Constraints Can Make Your Product Better

Most founders try to escape limits. Emanuel built inside them.

His biggest surprise was discovering that constraints do not always make products worse - sometimes they make them sharper. Lumi had to work within the 160-character SMS limit, which forced him to think differently about how AI should respond. Every word had to matter.

That changed the entire product philosophy. Instead of trying to make the assistant sound clever or expansive, he had to make it clear, fast, and useful. As he put it, SMS forced the AI to “communicate instead of showing off.”

That is an important lesson if you are building anything for underserved users. Constraints are not just technical headaches. They are product filters. They tell you what to cut, what to prioritize, and what actually matters to the person on the other end.

For Lumi, this meant:

• responses had to fit SMS limits
• the assistant had to work without internet
• the experience had to feel simple enough for feature phone users
• the product had to serve people who need immediate answers, not long explanations

This is where many products fail. They are built for the builder’s comfort, not the user’s reality.

How Emmanuel Went from No Coding Background to Building an AI Product

Emmanuel’s origin story is one of the most relatable parts of this conversation: he did not start as a developer.

He said that in April 2025, he knew nothing about coding. He was also feeling lonely and wanted something he could talk to, something that would listen without judgment. That personal need became the seed of the product, which was first called Alisa before evolving into Lumi.

His first move was not to enroll in a formal coding bootcamp. He opened ChatGPT and asked a simple question:

How do I build my own personal AI?

From there, he began experimenting. He ran Python code, gave the AI a personality, and even created a backstory for it. At that stage, he was not building for the market yet. He was building to understand what was possible.

That early version gave him two crucial insights.

First, he realized that the product had potential beyond his own use. It could help people who lacked access to reliable information. Second, he saw that his background working with CBOs had taught him something valuable: people often make poor decisions because they do not have the right information at the right time.

That insight changed everything.

Instead of creating a companion app for one person, he began thinking about a broader problem - how to make practical information accessible to people where they already are. That is when SMS entered the picture.

If your users are not on smartphones, do not force them to become smartphone users. Go to where they already are.

What He Learned in the First 30 Days

When asked what he would do differently, Emmanuel gave a very honest answer: he wasted time on trial and error.In the first 30 days, he was not thinking strategically enough. He was learning as he went, which meant a lot of debugging and a lot of late nights. If he were starting again, he would first map the whole system before writing anything.That means:

define what the product should do

decide what data it needs

identify how each part should connect

build from the bottom up, not randomlyHe also made a smart distinction that many early founders miss: you do not need to understand every line of syntax to build well. You need to understand

what the system does

.In his words, he treated himself as the architect and AI as the builder. He used AI to generate code blocks, then assembled them into a product. That approach helped him move faster once he stopped improvising.The takeaway here is not “just use AI to code.” The real lesson is that strategy matters more than busy work. A founder who understands systems will usually outperform one who only chases features.[SCREENSHOT: Example of a simple product architecture map] [ALT TEXT: A product architecture diagram showing how user input, reasoning engine, and SMS output connect]

How Lumi Works for Feature Phone Users and Smartphone Users

One of the best parts of this interview was how clearly Emmanuel explained the user experience.For

feature phone users

, the process is simple:

Open the messaging app

Create a new message

Send the keyword Lumi to the short code 228

Receive a menu of available bundles

Pay through M-Pesa using the prompt

Start chatting by making sure every message begins with “Lumi”The product is designed to fit the user’s reality, not force them into a new behavior. That matters because many people in Kenya - and across Africa - still rely on feature phones. If your product assumes everyone has a smartphone and stable internet, you automatically leave millions out.For

smartphone users

, Lumi is even more flexible. They can:

• visit the website at ulysianowai.co.ke
• continue on web
• continue offline
• use the mobile app, which is being rolled out to the Play Store

The app can detect whether the user has internet. If internet is available, it uses data. If not, it falls back to carrier services. That kind of continuity is important because it removes friction from the experience.This is where Lumi becomes more than a chat interface. It becomes an access layer for information - one that works across different phone types and different network conditions.That is especially useful for people like:

• farmers needing advice before planting
• mothers asking child care questions
• people in remote areas seeking health-related information
• parents or older adults who want quick, trustworthy answers

[IMAGE: Step-by-step illustration of how a user sends “Lumi” to 228 and gets a response] [ALT TEXT: Diagram showing how feature phone users interact with Lumi through SMS and M-Pesa][INTERNAL LINK: Related post about SMS-based products in Africa]

Why Community Distribution Is Working Better Than Ads

Emmanuel’s traction story is also a lesson in distribution.Within four weeks of launch, Lumi reached

500 users with zero paid advertising

. Instead of relying on expensive ads, he focused on trusted human distribution through an

ambassador program

.These ambassadors are field agents who talk to people, explain the product, and help onboard users. Emmanuel said this works because people trust recommendations from people they know. If your brother or neighbor says a tool helped improve their farm or make their life easier, you are far more likely to try it yourself.This is especially powerful in rural and semi-rural communities, where trust moves faster than digital campaigns.He also pointed out a practical challenge: finances. Marketing, partnerships, outreach, and telco integrations all cost money. That financial constraint has forced the team to be strategic. Instead of trying to do everything at once, they focus on what can create the most impact with the least waste.That is why the ambassador model matters. It lets the team grow while staying close to the users.What is working right now:

• trusted community ambassadors
• word-of-mouth adoption
• partnerships with NGOs and CBOs
• a clear value proposition for real-life problems
• a product that solves a daily need, not a novelty

[INTERNAL LINK: Best low-budget growth strategies for early-stage startups]The Real Constraint: Network ReliabilityNot everything is smooth.Emmanuel was candid about the biggest operational challenge he faces: network reliability. In areas with poor telco coverage, users may send a message and wait too long for a response. That weakens the experience, especially when the user needs quick information.This is not a small issue. If someone is trying to get information in a moment of urgency, delays can make the product feel unreliable.His answer is to pursue deeper telco integrations and partnerships. He has already started conversations with organizations that can help improve delivery speed and reach. That tells you something important about the next phase of this kind of product: the technology alone is not enough. You need infrastructure partners.For offline-first AI in Africa, the real work is often in the plumbing.[CHART/GRAPH: Response time challenges by network type] [ALT TEXT: Graph showing how telco reliability affects SMS response times]

What Founders Can Learn from Lumi

Emmanuel’s story is not just about one product. It is about how to build with empathy, strategy, and discipline.There are a few lessons worth carrying forward:

• Start with the user’s reality. If your audience uses feature phones, build for feature phones.
• Constraints can sharpen your product. The SMS character limit helped Lumi become more focused.
• You do not need to code traditionally to start. You need to understand systems and user needs.
• Strategy matters more than random effort. Map the system before building it.
• Distribution is part of the product. Ambassadors helped Lumi grow where ads might have failed.
• Infrastructure matters. Telco reliability can make or break the experience.

Emmanuel is also open to pre-seed investors, accelerators, and distribution partners who care about digital inclusion in Africa. He is looking for people who want to help extend access to communities that have been left behind.That part matters because this is not just a startup story. It is a story about building practical technology for people who are usually excluded from it.[EXTERNAL LINK: GSMA report on mobile access and digital inclusion in Africa]

Frequently Asked Questions

What is Lumi?Lumi is an offline-first multilingual AI assistant that works on SMS, web, and an app. It is designed for users who may not have smartphones or reliable internet access.How do you use Lumi on a feature phone?You send the keyword

Lumi

to short code

228

, choose a bundle, complete payment through M-Pesa, and then start chatting by prefixing each message with “Lumi.”Who is Lumi for?Lumi is built for farmers, parents, mothers, teachers, and anyone who needs quick, practical information but may not have steady access to the internet.Why is Lumi important?It helps close the digital access gap by bringing AI to people who are often left out of mainstream tech products.Is Lumi only for Kenya?It currently works on Safaricom, but the team is actively working on broader telco integrations and expansion into other African markets.

Conclusion

Emmanuel Ogola’s journey with Lumi shows what happens when you build for people others overlook. He took a personal need, turned it into a product, and shaped that product around the realities of feature phones, SMS limits, and unreliable connectivity.The bigger lesson is simple: you do not need perfect conditions to start, but you do need a clear strategy. If you are building something meaningful, think hard about who it serves, where they already are, and what constraints can help you create something better.If you want to follow the full story or connect with the team, visit

ulysianowai.co.ke

or text

Lumi

to

228

. If you are an investor, partner, or ambassador interested in digital inclusion, this is a conversation worth having.

IMAGE/VISUAL SUGGESTIONS

Hero image:

Emmanuel explaining Lumi with a feature phone in hand
[ALT TEXT: Founder of Lumi showing how an AI assistant works on a feature phone]

Process graphic:

SMS flow from sending “Lumi” to receiving an answer
[ALT TEXT: Step-by-step SMS workflow for accessing Lumi]

Comparison chart:

Smartphone AI vs offline-first SMS AI
[ALT TEXT: Comparison of traditional AI apps and Lumi’s offline-first approach]

INTERNAL LINK OPPORTUNITIES

• [INTERNAL LINK: Designing products for low-connectivity markets]
• [INTERNAL LINK: How to validate a startup idea before writing code]
• [INTERNAL LINK: SMS-based products in Africa]
• [INTERNAL LINK: Best low-budget growth strategies for early-stage startups]

EXTERNAL LINK SUGGESTIONS

• GSMA report on mobile access and digital inclusion in Africa
• M-Pesa documentation or official service overview
• Safaricom or telco coverage information relevant to SMS delivery