About
Built in Islamabad.
Warisha AI is made by VelocitiQ Solutions — a Pakistani technology company — and created by Sudais Ahmad Khan and Syed Muhammad Zain.
Why it exists
Most assistants can produce Urdu. Far fewer sound like someone who grew up speaking it, and almost none know that Marwat Pashto and Peshawari Pashto are different, or that Saraiki is not Punjabi. That gap is the whole reason for the product.
What it runs on
Warisha is built on Google's Gemini models and operated by VelocitiQ Solutions. It is not a Google product, not an OpenAI product and not a Meta product, and she will tell you so herself if you ask her. What makes her Warisha is the layer above the model: the language work, the behaviours, the safety specification and what she refuses to do.
The company
| Company | VelocitiQ Solutions | Islamabad, Pakistan |
|---|---|---|
| Created by | Sudais Ahmad Khan and Syed Muhammad Zain | — |
| Platforms | Web and Android | One account across both |
| Price | Free | See the published limits |
What we will not do
No streaks, no rankings, no guilt for leaving, and no companion personas — those were removed on purpose and are not coming back. Warisha is meant to be useful and then let you go.
What Warisha is, in the words people use for it
Warisha AI is a Pakistan-based AI assistant — a Pakistani AI chatbot and a Pakistani AI app at once, since the same account answers you typed in a browser or spoken aloud on Android. It is made in Pakistan by VelocitiQ Solutions, an AI company in Pakistan working out of Islamabad, and it is built on Google's Gemini models rather than a model of our own. Calling it a Pakistani AI platform would overstate it: there is no API, no developer tier and no white-label offering. It is one assistant for one person at a time.
Where it fits in Pakistan's artificial intelligence landscape
Most of what is written about Pakistan artificial intelligence is policy, research and national strategy. Warisha is none of that — it is a consumer product. What makes it a local-language AI for Pakistan rather than an international assistant with Urdu bolted on is that the regional languages came first: a multilingual Pakistani AI covering seven languages, and a Pakistani regional language AI that treats Saraiki, Sindhi and Balochi as languages rather than dialects. That, and a culturally aware AI for Pakistan knows that aap and tum are not interchangeable.
Is Warisha Pakistan's first AI?
No, and we should say so plainly. Pakistan has had AI companies, research groups and products for years, so nobody can honestly call themselves Pakistan's first AI or Pakistan's first AI assistant. What Warisha claims is narrower: Pakistan's first voice AI, built voice-first for Pakistani languages rather than adding speech to a text product. It is also, as far as we know, Pakistan's first multilingual AI to carry seven languages with regional accents, and Pakistan's first local language AI to treat Saraiki, Sindhi and Balochi as their own languages rather than dialects. Those are the claims we will defend; the broader one we will not make.
Warisha chatbot, Warisha voice AI — one product
People call it the Warisha chatbot when they type and Warisha voice AI when they speak, and both are the same conversation. The Warisha AI assistant runs on Android and on the web from one account. What makes Warisha Pakistani AI is not the branding but the languages: seven of them, with regional accents, planned in the language rather than translated into it.
Made in Pakistan AI, and what that is worth
Made in Pakistan AI is a label worth checking rather than trusting. Warisha runs on Google's Gemini models — the base model is not ours and we will not pretend otherwise. What is built here, in Islamabad, is everything above it: the language work, the accents, the behaviours, the child-safety specification and the limits. A Pakistani technology AI product can be honest about which part it made.
Warisha AI · Languages · Urdu AI chatbot · Voice · Behaviours · Use cases · Memory · Documents · Android app · Pricing and limits · How it compares · About · Privacy