Why We Teach AI Instead of Pretending It Is Not Coming

Let us be direct about this. We did not add AI to our curriculum because it is trendy. We added it because the job market demanded it, and pretending otherwise would have been dishonest to our students.

In 2024, AI coding tools crossed from early adopter curiosity to mainstream professional tool. GitHub reported that Copilot was being used by millions of developers. Companies started requiring AI tool familiarity in job descriptions. Development teams that adopted AI-assisted workflows shipped faster than teams that did not. The productivity difference was not marginal. It was significant enough that companies noticed and adjusted their expectations.

By 2026, a developer who does not use AI tools is like a developer who does not use Stack Overflow, or Google. You can technically function without it, but you are slower than everyone who does use it, and employers notice. When a team using Copilot and Claude delivers features in half the time, the team that refuses AI tools does not look principled. It looks slow.

We run a bootcamp in Kenya. Our graduates compete for jobs not just with other Kenyan developers, but with developers worldwide (especially for remote roles). Sending them into that market without AI skills would be like sending a boxer into the ring with one hand tied behind their back. We are not going to do that because the tools make us uncomfortable. Our comfort is not the point. Our graduates' employability is.

When we say we teach AI, we mean something specific. Not a vague "we encourage students to try ChatGPT." Here is what is actually in the curriculum.

AI-assisted development workflow. Students learn to use GitHub Copilot and Claude as coding partners from early in the program. This is not a one-time demo. It is integrated into daily work. When students build projects, they use AI tools to generate boilerplate, debug errors, explore unfamiliar APIs, and write tests. They learn which tasks AI handles well (repetitive patterns, documentation, simple functions) and which tasks require human judgment (architecture decisions, complex business logic, security-critical code).

Prompt engineering for developers. Writing a good prompt is a skill. Students learn to provide context, specify constraints, request explanations, and iterate on AI outputs. They learn that "write me a login page" produces generic garbage, while "generate a Next.js login component using Supabase Auth with email/password, include error handling, TypeScript types, and Tailwind CSS styling" produces something usable. The difference between a productive AI user and a frustrated one is prompt quality.

Critical evaluation of AI output. This is the most important AI skill we teach, and it is the one most programs skip. AI tools produce code that looks correct but often contains subtle bugs, security vulnerabilities, or architectural problems. Students learn to review AI-generated code the same way they would review a colleague's pull request: checking logic, testing edge cases, and questioning assumptions. Using AI without this critical layer is dangerous. We teach both the accelerator and the brake.

Building AI features. Beyond using AI tools for coding, students learn to build products that include AI capabilities. This means working with embeddings, vector databases, and RAG (Retrieval-Augmented Generation) to create features like AI-powered search, document summarization, and conversational interfaces. These are not theoretical exercises. Students build real features that users can interact with.

Understanding AI limitations. Students learn what AI cannot do. It cannot reliably handle novel architectural decisions. It produces confidently wrong answers. It can introduce security vulnerabilities if you accept its output uncritically. It is biased in ways that reflect its training data. Teaching AI without teaching its limitations would be irresponsible.

Some people worry that teaching AI tools undermines the learning of fundamentals. We see it the opposite way. AI tools are only useful if you understand what you are asking them to do. A developer who cannot read code cannot evaluate AI-generated code. A developer who does not understand databases cannot prompt AI to design a good schema. A developer who does not know React cannot use Copilot effectively in a React project.

AI is a multiplier. It multiplies what you already know. If you know nothing, the multiplication is zero. If you understand the fundamentals of programming, data structures, HTTP, databases, and frontend frameworks, AI tools make you dramatically faster and more capable. That is the sequence: learn the fundamentals, then add AI as an amplifier.

In our 26-week curriculum, the first several weeks are fundamentals without heavy AI reliance. Students learn JavaScript, then TypeScript, then React, understanding what they are doing and why. AI tools are introduced early but in a supporting role: "use Copilot to speed up typing this function you already understand." As students advance, the AI role grows: "use Claude to help you architect this feature, then evaluate whether its suggestion makes sense." By the end of the program, students use AI fluently but can also code without it when needed.

We test both capabilities. Can you use AI effectively? And can you code without it? Employers need both. Sometimes the Wi-Fi is down. Sometimes you are working on proprietary code that cannot be sent to an external AI. Sometimes the AI gives you garbage and you need to write the solution yourself. A developer who can only code with AI is fragile. A developer who can code with or without AI, and knows when each approach is appropriate, is resilient.

Some coding schools have taken the opposite approach. They ban AI tools during the program, arguing that students need to learn to code "the real way" first. We understand the impulse, but we think it is wrong.

The argument for banning sounds reasonable: if students rely on AI from day one, they never develop independent problem-solving skills. They become dependent on tools that do the thinking for them. They graduate unable to write code without Copilot autocompleting every line.

The problem with this argument is that it conflates using AI with depending on AI. We teach students to use AI strategically, not to abdicate their thinking to it. There is a big difference between "I asked Claude to write my entire project" and "I wrote the architecture myself, used Copilot for the boilerplate, and asked Claude to help me debug a tricky async issue." The first is dependency. The second is professional efficiency.

Banning AI tools also creates a false learning environment. When students graduate and start working, their employer will expect them to use AI tools. If they spent 6 months learning to code without AI and then suddenly need to integrate AI into their workflow, they lose time adapting. They also lose the months of practice that they could have been accumulating during the bootcamp.

The calculator analogy holds. When calculators were new, some maths teachers banned them. Students needed to do arithmetic by hand "the real way." That was understandable in the transition period. But eventually, the profession moved on. Nobody argues today that engineers should not use calculators. We teach the mathematical concepts and also teach how to use the tools that make those concepts practical. AI tools are the calculators of software development. Teach the concepts. Teach the tools. Both, not one or the other.

There is a reason we care about this more than most bootcamps, and it has to do with our specific context as an African-focused program.

African developers have historically competed at a disadvantage in the global market. Not because of talent (the talent is here) but because of structural factors: slower internet, less access to mentorship, fewer professional networks, and less exposure to production-grade codebases. These gaps make it harder for African developers to move as fast as their peers in Silicon Valley or London or Berlin, even when their raw ability is equal or greater.

AI tools compress those advantages. When a Kenyan developer uses Copilot, they get the same autocomplete suggestions as a developer in San Francisco. When they use Claude to explore an unfamiliar API, they get the same quality of explanation. AI tools do not care where you are sitting. They deliver the same multiplier to anyone who knows how to use them well.

This is a genuine equalizer. An African developer who masters AI-assisted development is more competitive for remote roles than an equally skilled developer anywhere who does not use AI tools. The speed advantage that AI provides is location-independent. Our graduates in Nairobi, Mombasa, Kisumu, and across Kenya should have that advantage from day one of their careers.

We also teach our students to build AI features because the African market needs them. Customer support chatbots for Kenyan businesses. AI-powered search for local content platforms. Intelligent document processing for organizations drowning in paperwork. These are real needs, and developers who can build AI-powered solutions for the African market are in demand. We are not teaching AI because Silicon Valley says it is cool. We are teaching it because Kenyan businesses need developers who can build it.

Theory is fine. Here is what our students actually build.

AI-powered search. Students build search features that go beyond keyword matching. Using embeddings and vector databases, they create search systems that understand meaning. A user searching for "how to send money" finds results about M-Pesa, EFTs, and mobile wallets, even if those exact words are not in the content. This is RAG (Retrieval-Augmented Generation) in practice, and it is one of the most in-demand AI features in production applications right now.

Conversational interfaces. Students build chat-style interfaces where users can ask questions in natural language and get useful answers. This is not a generic ChatGPT wrapper. It is a focused, purpose-built assistant backed by specific data (a product catalog, a knowledge base, a support documentation set). Building this teaches API integration, prompt engineering, context management, and UX design for AI interactions.

AI-assisted content tools. Students build applications that help users generate, edit, or organize content using AI. Think of tools that help small businesses write product descriptions, generate social media posts, or summarize long documents. These projects combine frontend skills (building the interface), backend skills (managing the AI API calls), and product thinking (what is actually useful to the end user).

Intelligent form processing. Students build systems that take unstructured input (a paragraph of text, a scanned document, a voice recording) and extract structured data from it. This is practical and immediately useful for businesses that deal with invoices, applications, or any paperwork-heavy process.

Every one of these projects involves real AI APIs, real user interfaces, real database operations, and real deployment. They are portfolio pieces that demonstrate not just that the student can code, but that they can build modern, AI-powered products.

If you are considering a coding bootcamp in 2026, the AI question is not optional. You need to ask every program you are evaluating: "Do you teach AI-assisted development? How?" The answer tells you whether the program is preparing you for the market as it exists today or the market as it existed three years ago.

At McTaba Labs, our 26-week Software and AI Engineering program integrates AI throughout the curriculum. You learn TypeScript, React, Next.js, Node.js, PostgreSQL, and Supabase. And you learn to use AI tools from early in the program, building up to full AI-feature development by the end. The curriculum costs KES 100,000, runs part-time (so you can keep working), and is updated continuously.

We are not the only program worth considering. But we are one of the few in Kenya that treats AI as a core curriculum component rather than a marketing bullet point. If you compare our syllabus against any other program, the AI integration will be one of the most visible differences.

The broader point is this: the developer job market in 2026 rewards people who can use every tool available to them. AI is the most powerful new tool to arrive in a generation. The question is not whether you should learn it. The question is whether you learn it now, as part of a structured program, or later, on your own, while your peers who learned it earlier are already putting it to work. We think now is better. That is why we teach it.

If you want to start small before committing to the full program, our Tech Foundations course (KES 2,999) introduces the basics and gives you a feel for how we teach. No pressure. Try it, see if the approach works for you, and go from there.