Chisimba 26: How We Got Here

Chisimba, renewed: building a modern platform for learning and publishing

Chisimba began with an ambitious idea: African universities could build useful Free Software together while developing their own software-engineering capacity. That history, rooted in the African Virtual Open Initiatives and Resources (AVOIR) and NetTel@Africa collaboration, remains worth building on.

The work now under way brings Chisimba into a different technical and educational world. Modernisation means updating the software, but it also means rethinking how people create courses, publish ideas, and navigate the application. It means making room for AI while keeping human judgement central.

The latest update to Chisimba.com is a practical step in that direction. We now have a rebuilt publishing workflow to use for these updates—and an opportunity to explain the thinking behind it.

Keep the modular foundation, improve what modules share

Chisimba has long kept its core framework separate from the modules that provide specific capabilities. That remains a useful foundation. The modernised site runs on PHP 8.5, while the application continues to build on its modular, model–view–controller (MVC) architecture.

The architectural work increasingly concerns what modules should share. Identity, permissions, media selection, author biographies, classification and AI access should have clear homes. A module should use those services rather than invent its own version.

Categories and tags are a good example. We have introduced a shared classification service and connected it to the blog. This gives other modules a foundation to adopt, rather than leaving classification trapped inside one publishing tool.

Language is part of the architecture too. Chisimba separates concepts from the words a particular site uses for them. A blog can become “News and Views”; a post can become an “article” or “news item”. The underlying functionality stays consistent while the terminology fits its purpose.

Publishing with building blocks

A large text box is useful, but it should not be the whole publishing experience. The rebuilt blog is a good example of how Chisimba 26 uses a reusable content-composition service. Authors can combine text, image-and-text layouts, heroes, video and sliders, then arrange those blocks into a post. One block is edited at a time while the others show previews. Dragging and button controls let you change the order.

A featured image belongs to the post itself, independently of images within its content. Post listings use cards, with a consistent image area, an excerpt and a clear route into the article. Categories, tags, archives and reusable sidebar blocks help readers explore beyond the page they first encounter.

The aim is to improve a shared builder that can serve multiple modules. It is a foundation for wider reuse; it does not mean every older page-building interface has already been converted.

Let real work expose the awkward parts

Some of the most useful design feedback has come from building a course about the grasses of the South African savanna. Planning that course with the Active Knowledge Map exposed practical problems: crowded branches, overlapping nodes, unclear ordering, and controls that used too much of the available space. Those are consequential problems when the map is both a working outline and something to share with another organisation.

The publishing work has followed the same pattern. Using the interface quickly reveals whether a search field is too wide, a button is the wrong height, or Help occupies an unnecessary row. These details influence how comfortably someone can work.

Shared skin primitives help make those improvements consistent. A clearer button, card or focus treatment should become something other parts of Chisimba can reuse.

AI should assist accountable people

AI is part of this modernisation, both as a development aid and as a capability that Chisimba modules can use. Within the application, a shared AI service separates provider execution from the educational or publishing task. The calling module supplies its instructions and expected structure; the service manages the provider interaction and records request metadata. This gives us a common place to manage access and visibility into AI use.

That architecture helps, but it does not make an answer correct, fair or appropriate. Responsible use still requires people to decide what a tool should do, what information it should receive, and whether its output is fit for purpose.

For educational content, that means checking facts, relevance and the quality of assessment questions. For development, it means inspecting changes, testing behaviour and protecting existing data. Privacy, cost, and the ability to understand failures also factor into those decisions.

The working principle is straightforward: AI can help produce a draft or explore an approach. People remain responsible for what they teach, publish, and deploy.

A platform to use and improve

The latest release also brings reusable author biographies and course marketing pages into the broader picture. They connect the people behind a course with the information prospective learners need before joining.

There is more to do. Moving WordPress content into Chisimba is still ahead, and using these tools will uncover further improvements. This release lets us put the new publishing workflow to work on Chisimba.com itself.

Chisimba’s collaborative origins still provide a useful direction: build openly, share what you can, and let practical use guide the next improvement.

This post was drafted with AI assistance. The featured illustration was generated with AI. I used Codex extensively to carry out this Chisimba modernisation.