Why We Chose WhatsApp

When we started building Mizar, the first question wasn't "what AI model should we use?" — it was "where should the AI live?"

The answer, for us, was obvious. But it's worth explaining why.

The app fatigue problem

There are hundreds of AI apps available today. ChatGPT, Claude, Gemini, Perplexity, Copilot — the list goes on. They're all impressive. They're all powerful. And most people have tried at least one.

But here's the thing: trying an AI app and actually using it daily are two very different things. Most people download an AI app, use it a few times, and then forget about it. It becomes another icon on a screen already full of apps competing for attention.

The friction of opening a separate app, switching contexts, and adapting to a new interface is enough to prevent AI from becoming a daily habit for most people.

Meeting people where they are

WhatsApp has over 2 billion active users. In many parts of the world — Latin America, Europe, Africa, South Asia — it's not just a messaging app, it's the primary communication tool. People spend hours in WhatsApp every day.

By building Mizar on WhatsApp, we eliminated the biggest barrier to AI adoption: the need to go somewhere new. There's no app to download, no interface to learn, no new habit to build. You just message Mizar like you'd message a friend.

The power of conversational UI

WhatsApp's chat interface is the most intuitive UI in the world. Everyone knows how to send a message. Everyone knows how to send a photo. Everyone knows how to send a voice note.

This means Mizar is immediately accessible to people who might never use a traditional AI app — your parents, your grandparents, people who aren't "tech-savvy." If they can send a WhatsApp message, they can use AI.

And because WhatsApp supports rich media — images, documents, voice messages, locations — Mizar can accept and respond with all of these formats natively.

Always with you

Your phone is always in your pocket. WhatsApp is always open. This means Mizar is always accessible, in every context:

• •Commuting: Ask questions hands-free with voice messages
• •Shopping: Send a photo of a product label for instant analysis
• •Cooking: Ask for recipe substitutions while your hands are messy
• •Traveling: Get real-time translations and local recommendations
• •Working: Quick research without switching away from your workflow

This ambient accessibility is something no standalone app can replicate.

The challenges

Building on WhatsApp isn't without tradeoffs. The platform has limitations:

No rich UI components. We can't show buttons, carousels, or interactive elements the way a native app could. Everything is text, images, and documents.

Message formatting is limited. We work with bold, italic, monospace, and basic lists. No markdown tables, no syntax highlighting.

Rate limits exist. WhatsApp's Business API has sending limits that we need to work within.

No real-time streaming. Unlike web-based AI chatbots that can stream responses token by token, WhatsApp delivers complete messages. We optimize for fast, complete responses.

We've embraced these constraints rather than fighting them. They force us to be concise, clear, and direct — which, honestly, makes for a better AI assistant.

The big picture

We believe the future of AI isn't in standalone apps — it's embedded in the tools people already use. Email, messaging, calendars, browsers. AI should be a layer that enhances everything, not a destination you have to visit.

WhatsApp is our starting point, but the vision is bigger. Mizar is your personal AI — and it should be available wherever you need it.

For now, though, the simplest version of that vision is already live: just send a WhatsApp message and see what happens.