Can You Build a Chatbot in Five Minutes? What I’m Building with Nzitsa Block Studio

I started Nzitsa AI Labs with a question: what would it take for more Zambians to build useful AI systems, rather than only consume tools built elsewhere?

The gap is easy to see. A student, teacher, founder, or small organisation can open a global AI tool and get an answer almost immediately. But imagine a lecturer who wants a focused study assistant for their own course: one that has a clear purpose, a familiar interface, a selected model, sensible limits, and a path to being reviewed before students use it. Turning that idea into a real tool involves far more than typing a prompt. It requires model access, usage controls, cost awareness, technical work, trust, and people who can operate the system responsibly.

That contrast is what motivates Nzitsa AI Labs and Nzitsa Nzeru.

The experience I want to make possible

I am building Block Studio around a straightforward challenge: could an approved builder start with a reviewed chat template, describe what the assistant should do, choose an AI model available to their account, shape the interface, and preview a ChatGPT-style chatbot in about five minutes?

The word Block matters here. A Block is a reusable AI application template, not a new foundation model. The initial workflow is designed around a reviewed chat-assistant template, visual editing, and desktop/mobile previews. The builder can change the product identity, choose from bounded appearance options, arrange trusted components, write the assistant guidance, and select an account-available model alias. The preview is intentionally local: it shows the experience without calling a model or spending quota.

This is the five-minute creation flow I am working toward. I am not claiming a completed, public end-to-end run has already been timed at five minutes. And I am definitely not saying that somebody can train a new model in five minutes. The promise is narrower and more useful: reduce the early distance between a locally relevant idea and a well-shaped chatbot experience, while keeping the serious work visible.

The harder part starts after the preview

A convincing preview is only the beginning. Before another person can rely on a chatbot, someone must decide who is allowed to use it, which model is available, how its usage is limited, how costs are handled, and whether its design and purpose meet the standard for publication.

That is why Nzitsa AI Labs is not simply a page that sends prompts to a model. It is being built as a governed platform for shared AI work. Nzitsa manages identity, permissions, model aliases, routing, quotas, cost records, approvals, and audit trails. Its repository also includes workflows for datasets, model experiments, collaboration, reviewed projects, and release evidence.

The underlying inference and remote engineering services can come from hosted providers. That is a deliberate decision. I do not need to claim that Nzitsa owns every GPU or every foundation model in order to build a locally governed service. The local control that matters today is deciding how the platform is run: who gets access, which options are safe for an account, how spending is constrained, how work is reviewed, and what kinds of products are allowed to reach users.

Some people describe this ambition as sovereign AI. I use that phrase carefully. For Nzitsa, it means local decision-making, local talent development, and accountable operation of the platform—not an inflated claim that the entire AI stack is locally owned.

Nzeru is the people side of the platform

Technology without a route for people to learn and contribute tends to reinforce the same dependency it is trying to solve. Nzitsa Nzeru is the human thread in this work. Its role is to grow practical confidence through the planned Know About AI seminars, mentorship, shared learning, and project participation.

AI Labs and Nzeru are designed to strengthen one another. Nzeru helps people move from curiosity into capability. AI Labs gives approved builders governed tools, collaborative engineering workflows, and opportunities to contribute to reviewed projects. That connection matters to me because a local AI ecosystem should create more than users; it should create people who can question, adapt, evaluate, and operate the systems their communities depend on.

The K80,000 EIZ Engineering Fund award announced on 10 April 2026 gives the initiative meaningful momentum under the theme of AI for Sustainable Development. AI Labs and Nzeru are products of that direction. The remaining grant funding is planned for the Know About AI seminar programme, so the platform work is accompanied by public learning and conversation rather than being built in isolation.

What is implemented—and what still needs proof

It is important to separate local implementation from public availability. Locally, the project has implemented the provider gateway, collaboration and project workflows, dataset lifecycle work, Block Studio’s reviewed registry and visual preview, and fine-tuning lifecycle controls. Those are the foundations that make a governed builder experience possible.

Provider-backed transformations, Space deployments, resource-group operations, and fine-tuning are not being presented as completed public services. They remain behind release controls until official accounts, billing ceilings, reviewed runtime bundles, live staging evidence, and release approval are in place. That boundary is not a weakness in the story; it is part of building responsibly when external services can cost money or affect other people.

An invitation to build the right things

I want to hear from builders who have a useful local problem, educators who can see where a focused assistant would genuinely help learners, researchers who want to evaluate ideas rigorously, and organisations that need an accountable route into AI experimentation.

You can follow the product work at ai.nzitsa.com, see the wider Nzitsa Platforms ecosystem, and explore the education work at Nzitsa Academy. If you want to collaborate, propose a practical use case, contribute to a seminar, or discuss a reviewed project, please get in touch.