Lagos, Nigeria — Building Physical AI

Physical AI for Africa.
Built from Lagos.

Building the infrastructure layer for Physical AI in the Global Majority, designed from the start for the power, bandwidth, and hardware realities the industry's defaults assume away. What works here makes AI work everywhere.

See what we're building ↓
The thesis

Physical AI is being designed for Western homes, smart cities, and enterprise facilities. The billions of people in Nigeria, India, Brazil, and the rest of the Global Majority are absent from those assumptions.

Our approach

Build practical systems that work inside the infrastructure people actually have: intermittent power, unstable internet, mid-tier hardware, messaging apps in place of broadband. Ship locally first. Expand outward.

The horizon

Each deployment, with consent, produces real-world vision-language data from environments almost entirely missing from current training sets: the infrastructure layer for physical AI in emerging markets.

For developing countries, "AI is likely to close the technology divide," offering a chance to narrow long-standing technology gaps through its accessibility and abundance.
— Jensen Huang, CEO of NVIDIA

We agree, with a caveat. Every previous technology wave arrived in Africa with assumptions that didn't hold: broadband, app stores, credit cards for subscriptions. AI is the first wave we can design for African constraints from the beginning, instead of retrofitting it after.

But the divide only closes if the infrastructure layer that makes adoption possible gets built. That's what WTMG Labs is: starting from a $35 Raspberry Pi and a WhatsApp message, scaling to the data and systems that power the continent's physical world.

The next wave of AI will be shaped by data from Lagos, Nairobi, and Accra, or it won't work there at all.

First product
Gateman-AI
Open Source

Chat with your cameras on WhatsApp.

A WhatsApp-native security and accessibility agent for homes and small businesses. Connects a camera in your environment to WhatsApp: no custom app to install, no separate cloud storage subscription. Runs on a $35 Raspberry Pi or any existing RTSP camera. All AI inference stays local. Nothing leaves your network.

View on GitHub →
"The gate was locked for over a year. We had no way to know when it was safe to be out."

A bakery manager near Yabatech Polytechnic in Lagos described the daily fear of not knowing who was outside during a period of repeated gun violence in the area. Armed forces guard the school. Civilians around it have nothing.

The Raspberry Pi powering Gateman-AI came back from Austin, Texas in my luggage. Two years at Eagle Eye Networks working with enterprise security camera infrastructure showed me exactly what it could do, and exactly who it could not reach.

Back in Lagos, that gap was impossible to ignore.