Making Manifestos Accessible: A WhatsApp Chatbot for Informed Voting

Election season in South Africa arrives like a weather system. Suddenly, posters bloom on lampposts. Faces you've never seen smile at you from every corner. And every political party releases a manifesto — a thick document explaining their policies, plans, and vision for the country.

These manifestos are meant to help. They're meant to inform. They're meant to give voters the tools to make real choices.

There's just one problem.

Almost nobody reads them.

The Uncomfortable Truth

Manifestos are dense. Packed with policy details, economic projections, legal language that makes your eyes slide off the page. Most people don't have hours to sit down and work through fifty pages of promises for every party on the ballot.

So they vote on vibes. On tribal loyalties. On what their parents voted, or what their friends are voting, or which candidate had the catchiest jingle.

I'm not judging. I get it. Life is busy. Information is scattered. And reading policy documents for fun isn't most people's idea of a good time.

But there's something sad about it, isn't there? Democracy asking us to choose, and most of us choosing half-blind.

I wanted to do something about that.

The Idea That Wouldn't Let Go

The concept surfaced during a conversation with a friend. We were talking about the upcoming elections — the usual mix of cynicism and hope that these conversations carry.

*What if you could just ask?*

Not wade through PDFs. Not decode political language. Just type a question and get a real answer.

"What does this party say about unemployment?" "What's their stance on education?" "How do they plan to address load shedding?"

Real questions. Actual policy positions. Delivered through the app already living in everyone's pocket: WhatsApp.

The idea lodged in my brain like a splinter. I couldn't stop thinking about it.

Why Rise Mzansi?

I needed a test case, and Rise Mzansi felt like the right fit.

They positioned themselves as a modern, fresh political movement — newer energy, different approach. If any party would be open to their manifesto being made accessible through a chatbot, it seemed like them.

This was a hobby project. No client. No contract. No revenue model. Just an idea I believed should exist, and enough skill to make it real.

Rise Mzansi's manifesto became the proof of concept.

The Technology Underneath

The approach was called Retrieval-Augmented Generation — RAG, if you want the shorthand.

Here's what that means, stripped of jargon: instead of training an entire AI model on the manifesto (expensive, time-consuming, probably overkill), RAG lets you keep the document as a knowledge base. When someone asks a question, the system searches through the manifesto, finds the relevant sections, and uses AI to turn that into a natural, conversational response.

I built it using:

• **WhatsApp** as the interface — meet people where they already are
• **Botpress** as the chatbot platform — it handled the retrieval, the chunking, the intelligence

The beautiful thing about this approach? Answers come from the source. The chatbot isn't guessing or generalising. It's pulling actual policy positions from the actual manifesto. If the information exists, you get it. If it doesn't, the bot says so.

Honesty built into the architecture.

The Experience of Using It

Simple and conversational. That was the whole point.

You message the WhatsApp number. Type your question in plain language — the way you'd ask a friend.

*"What does Rise Mzansi plan to do about crime?"*

The chatbot searches, finds the relevant sections, and responds with a clear summary of the party's position.

No PDFs to download. No websites to navigate. No legal language to decode. Just a conversation.

That's what accessibility looks like. Not making information available. Making it *reachable*.

The Hard Part Wasn't Technical

The trickiest challenge wasn't about code or APIs. It was about tone.

It would have been easy to build something stiff. A robotic Q&A that only responded to precisely-worded questions in precisely the right format. But that's not how people talk. They ask loosely. They use slang. They wander off-topic. They phrase things in ways that make perfect sense to them and zero sense to a database query.

I wanted the chatbot to feel like talking to someone helpful. Someone who knows the content deeply but doesn't make you feel stupid for asking.

At the same time — and this is crucial — I had to be careful about accuracy. This is political information. Getting it wrong could mislead voters. The chatbot needed to be helpful without overstepping.

So I built in honesty about limitations. When the manifesto doesn't cover something, the bot says so. *"I'm not sure about that"* or *"The manifesto doesn't address this specifically"* — these are valid, valuable responses.

Trust comes from acknowledging what you don't know.

Why This Matters to Me

This wasn't a paid project. Nobody asked for it. I built it because I believed it should exist.

That's the kind of work I want to represent. Technology that contributes to society.

Democracy works best when voters are informed. But information access isn't equal. Some people have time to research. Others are working three jobs. Some are comfortable navigating government websites and PDF documents. Others barely have data on their phones.

A WhatsApp chatbot levels the playing field, even just a little. It takes information that's technically public but practically unreachable and puts it in everyone's pocket.

That's the kind of tech I want to build. Not because it's impressive. Because it matters.

What's Next?

Right now, the project sleeps. The election passed. The immediate urgency faded.

But elections come around again. And next time, who knows?

Maybe this becomes a platform covering multiple parties. Maybe voters can compare positions across manifestos with a single question. Maybe political organisations see the value and want their own version.

The proof of concept exists. The technology works. It's waiting for the right moment to wake up.

The Bigger Picture

Not every project needs to be a business. Not every build needs a revenue model.

Sometimes you create something because you see a gap and you have the skills to fill it. Sometimes the reward is knowing that even one person made a more informed decision because of something you built.

This chatbot represents that philosophy. Civic tech — technology in service of citizenship.

And it's a reminder that the skills I've developed aren't just for commercial work. They can contribute to the kind of society I want to live in.

Maybe that's naive. Maybe it's idealistic. But I'd rather build something that might help than build nothing because I wasn't sure it would succeed.

---

Technical Summary

**Timeline**: Built a couple of months before the 2024 elections

**Type**: Personal/hobby project

**Status**: Shelved (potential revival for future elections)

**Key Value**: Technology that contributes to society and informed democracy

---

*Interested in civic tech or chatbot solutions? [Let's talk.]*