Working examples

I build custom software that runs your business operations.

I ship working products, not decks. Here are two you can try today: one for Jumia sellers, one for real estate agents.

Featured projects

Two tools, built from scratch.

Market intelligence · 32 Jumia sellers

StillWater

Market intelligence for Jumia Kenya. See what’s listed, compare prices and check how crowded a category is before you spend on stock. Includes an AI assistant that tells you, in plain words, what to test and what to skip.

Read the case study
Live demo · 11 paid real estate companies

WhatsApp Real Estate Lead Bot

Answers property enquiries on WhatsApp in seconds. Sends matching listings as photos with captions, then hands the lead to an available agent, with every lead logged in your CRM. Try it live.

Try it on WhatsApp ↗
Live demo
Case study 01

A WhatsApp assistant that answers property leads in seconds.

Most property leads arrive as a message on a phone. If the reply takes hours, the lead moves on to the next agent. This bot answers straight away, shows real listings with photos, and hands the buyer to a human agent as soon as they pick a property.

11paid real estate companies use this software

The use case

A real estate agency runs a Facebook ad for properties. A buyer taps the ad and lands in WhatsApp with a message already written. Today, someone on the team has to read it, find listings, send photos and decide who follows up. The bot does the first part for them, every time, at any hour.

How it works for the buyer

Six steps, from the ad to a human agent. Mock-ups with demo data.

STEP 01 · Tap the ad
NH
Nairobi HomesSponsored · demo page
New apartments and houses across Nairobi. Tap to see what’s available.
wa.meSee available properties
Chat on WhatsApp
👍 Like💬 Comment↗ Share

The buyer sees a property ad on Facebook and taps the WhatsApp button.

STEP 02 · WhatsApp opens
‹
NH
Nairobi HomesBusiness account
Chat with us to get started.
Hi, I’d like to see available properties in Nairobi.
➤

The first message is already written. The buyer only taps send.

STEP 03 · Pick an area
‹
NH
Nairobi HomesBusiness account
Hi, I’d like to see available properties in Nairobi.
Which area would you like to see? Pick a location and I’ll show you what’s available, with photos.
☰ Choose location
Westlands3 available · from KES 8.9M
Kilimani2 available · from KES 12.5M
Kileleshwa2 available · from KES 13.6M

A short list of locations, with how many properties are available in each.

STEP 04 · See the properties
‹
NH
Nairobi HomesBusiness account
Kilimani
🏠 Riverbend Heights, 2BR Apartment
📍 Kilimani · Apartment
💰 KES 12,500,000
🛏 2 bed · 🛁 2 bath · 105 m²
🗺 Around: Yaya Centre, Prestige Plaza
Choose this one

Each property arrives as a photo with the price, size, features and what is nearby.

STEP 05 · Choose one
‹
NH
Nairobi HomesBusiness account
Choose this one
Connecting you with an agent now. They’ll message you in a moment, and I’ve shared the details so you won’t need to repeat anything.

One tap. No typing and no reference codes to remember.

STEP 06 · The agent gets the lead
‹
AG
Lead alertsAgent’s WhatsApp
New lead from the WhatsApp bot (demo):

Name: Sample Buyer
Number: +254 7XX XXX XXX
Property: Riverbend Heights (KI-001), Kilimani, KES 12,500,000
Stated budget: not given

An available agent receives the buyer’s name, number and chosen property, ready to call.

Mock-ups of the real flow. The page name, buyer and listings are illustrative demo data.

The value

Fast first reply

The buyer gets an answer while they are still interested, not the next morning.

Agents talk to ready buyers

By the time an agent steps in, the buyer has already chosen a property and shared a budget.

Every lead is captured

Every conversation and every customer is logged automatically. Phone numbers are collected and saved to your CRM as it happens, at the first point of contact.

Connects to your tools

It works with the CRM or internal software your company already uses, so leads land where your team already works.

Custom built for each company

It is written as custom code around how your company works, so it fits your process instead of forcing you to change it.

Scales as you grow

The data behind it is built to grow, so more conversations, customers and listings don’t slow it down.

What’s behind it

It runs on the WhatsApp Business Platform, with a database behind it that holds the listings, photos, agents and every conversation. Messages are understood from plain text (“2 bed in Westlands under 15 million”), and the bot only ever shows properties that exist in the database. It does not invent listings. Because it is custom code, it connects to the CRM or internal software each company already uses, and every lead lands there as it happens. It can also run on the client’s own server, so lead and property data stays with the business. Message content still passes through WhatsApp’s own systems.

Used by 11 paid real estate companies. The demo you can try above uses fictional listings and photos, so no client inventory is shown.
Case study 02

Stop guessing what will sell on Jumia.

StillWater is a market-intelligence platform for Jumia Kenya. Before someone spends money on stock, it lets them check what is listed, compare seller prices, track price history, see how crowded a category is, and see which brands keep showing up. An AI assistant answers plain-language questions about the data, for people who do not want to read tables.

32Jumia sellers with real stores use StillWater
7 monthsof steady, consistent use

The problem it solves

01

Importing blind

You order 100 pieces because the product looks popular. Then you find a crowd of sellers offering the same thing, and your cash is stuck in stock.

02

Pricing by guesswork

Price too high and buyers pass. Price too low and Jumia fees eat the margin.

03

Decisions from group tips

A tip in a WhatsApp group is not the whole market.

One bad stock order can tie up a small business’s cash for months.

How the data works

  1. Collect. It tracks what is listed on Jumia Kenya: products, sellers, prices, categories, brands, ratings and reviews.
  2. Record changes over time. Each observed price is stored, building a price history.
  3. Filter and organise. Raw listings are cleaned and grouped by category, brand and seller, so they can be searched and compared.
  4. Measure the market. For each category it shows how many listings exist and how crowded it is.
  5. Answer questions. The AI assistant reads the same data you can see and answers in simple language.
StillWater shows what is listed and observed on Jumia. Listing position, ratings, reviews and changes are clues about what moves. They are not proof of orders or profit.

How decisions get made with it

Scenarios use made-up products. Numbers are illustrative.

Before importing

A first-time importer checks a product idea. The category shows many sellers at similar prices, so they hold off or pick a variation with fewer competitors.

Pricing

A store owner sees the range of seller prices and the price history, then prices inside the range in a way that still covers Jumia fees.

Choosing what to test

A side-hustler looks for categories with moderate competition and rising listings, and places a small first order.

Defending a position

An established vendor watches competitors’ prices and spots undercutting early.

The value

Know before you spend

Moves the decision from guessing to checking, before any stock is ordered. This is the main benefit.

Protect cash

Checking crowded categories first lowers the risk of money sitting in slow stock.

Protect margin

Seeing real seller prices helps you set a price that covers fees and still competes.

Save hours

No more scrolling Jumia category by category. One view with filters and history.

Easy for everyone

The AI chat gives a plain answer to people who do not want to read tables: what to sell, what to avoid, and why.

Spot movement early

Price history shows where the market is shifting.

Built for small sellers

For the Kenyan seller, from the first-time importer to an established Jumia store.

StillWater shows the market clearly. It does not promise sales.

Inside the platform

Mock-ups of the interface. Illustrative data.

OverviewIllustrative data
Products•••
Scanned•••
Analysed•••
Kitchenbusy
Phone accessoriesvery busy
Garden toolsroom

Overview: the scale of the data, with category and brand panels.

Market checkIllustrative data

Worth testing

Sample item AModerate competition, steady listings
Sample item BFew sellers, wide price range

Think twice

Sample item CMany sellers, tight prices
Sample item DHeavy undercutting

Indicators, not guarantees.

Worth testing versus think twice, each with a short reason.

Ask StillWaterIllustrative data
What should I sell this month in kitchen appliances?
Based on the listings I can see:
  • Test: items with fewer sellers and a wide price range.
  • Skip: categories where many sellers sit at close prices, as that would likely mean a price fight.

The AI assistant answers in plain words, based on the data.

How I work

Four steps, no jargon.

STEP 01

Understand your business

We talk through how work actually flows today and where it slows down.

STEP 02

Build a working version

You get something real to click, not a slide deck.

STEP 03

You test it on real work

Your team uses it on live tasks and tells me what’s off.

STEP 04

Refine and hand over

We fix what you found, then hand it over. Hosting can be on your own server, so your business data stays with you.

What I can build for your business

Tools shaped around one business.

Order capture on WhatsApp

Customers order in chat. You get a clean list.

Inventory tracking

Know what is in stock and what is running low.

Quotes and invoices

Send them in minutes, not hours.

Client ordering portal

Regular customers reorder on their own.

Delivery routing

Plan drops in a sensible order.

Payment reminders

Polite nudges go out without you chasing.

Sales dashboards

See what is selling in one view.

A private AI assistant

Ask questions about your own data.

About

Griffin Obungah

I’m a software builder based in Nairobi. I taught myself to code, and I hold a BSc in International Business and Trade, with hands-on operations and marketing experience. That means I look at your business first and the software second. I build tools from scratch for one business, and I can run them on your own server.