Software, AI agents and automation — built for real operations, not for demos.
Three systems from our work: one for a client, two of our own.
status as of Sep 2026A parcel and errand delivery platform connecting customers with a managed courier fleet through delivery requests, automated dispatch, live tracking and delivery operations.
Including dispatch, live tracking, multi-stop delivery and a driver wallet.
A multi-vendor food-delivery platform connecting customers, restaurants, drivers and operations teams through ordering, dispatch, delivery tracking and restaurant operations.
Including dispatch, live tracking and a restaurant POS.
A student-services platform supporting the process of studying in Rwanda through university services, document equivalence, housing, currency exchange and integrated support.
Including chat, document equivalence and a news and social feed.
The full work page adds our other products.
See all workSoftware that runs the operation, AI that understands and decides, and automation that connects and acts.
The system itself: the apps, backends and business logic your operation runs on.
handoff / your software's data feeds the agent
The layer that reads your data, reasons over it and takes part in the work.
handoff / the agent's decision triggers the workflow
The layer that acts: moving work between people, systems and the tools you already use.
Not sure which layer you need? Start with the problem, and we will work out the system with you.
Start a projectWhere we test intelligent systems, automation and decision tools before we treat them as products.
status as of Sep 2026SudaFood's backend already flags orders a restaurant leaves pending too long, and an operator then had to contact the restaurant by hand. This prototype tests automating that follow-up without giving a model unrestricted control. It acts on the backend's pending_late signal; it does not detect delays itself.
After a manual start it finds eligible orders, checks them, decides whether to send a reminder, sends it on WhatsApp and logs the send. It cannot change orders, dispatch drivers, contact customers or run continuously on its own.
How can we evaluate a startup opportunity without mistaking desk research for market validation?
Idea Scout turns an idea into hypotheses and research questions, researches the most critical one on the live web, classifies the evidence and returns decision support with a next experiment. It cannot approve an MVP, contact customers or run field validation.
BUILD_CANDIDATE means ready for founder review and field validation. It is not an approval to build.
The Lab page lists each experiment's evidence and where it stands.
Explore the LabAn African technology company building software, AI and automation, for clients and for the products we run ourselves.
how we work / build · launch · scale
Engineer the whole system, not only the screens.
Put it into real use and learn from how it operates.
Improve, automate and extend what proves valuable.
A core team owns every project end to end. Specialists join around the technical needs each project actually has.
We started MOTNIX because we wanted to build, operate and learn from real systems — not just deliver screens.
Every project on this site is labelled for exactly what it is: client work, venture, prototype or simulation.