Leveraging GenAI to supercharge your Muslim startup

الحمد لله رب العالمين، والصلاة والسلام على رسول الله

Salams friends,

This post is a collection of notes I took while listening to Dr. Waleed Kadous at the Muslim Biz Summit. Be sure to follow them for video content of these sessions, and keep an eye out for the next summit!

Dr. Waleed Kadous is the Engineering Lead for Generative AI at Canva. Previously, he served as the Engineering Strategy Lead at Uber where he championed machine learning, and as a Principal Engineer at Google on the Maps and Android Location team, responsible for the ubiquitous blue dot we all use today.

Rather than providing another "Introduction to AI" (of which there are many), this post focuses on how generative AI can be leveraged by startups, both Muslim and non-Muslim, and its emerging role in Islamic scholarship, including Dr. Kadous's innovative tool, Ansari.Chat.

So bismillah, let’s get into it!

Table of Contents

• Embracing AI

• 6 Key Areas of Implementation

  • 6. Leveraging Your Data

Embracing AI

Dr. Kadous emphasizes that cultivating a culture of AI adoption is paramount.

The common fear of AI replacing jobs misses the point - the real distinction will be between humans who embrace and leverage AI and those who don't. This is illustrated by a GitHub Copilot study showing that developers using the tool completed tasks 55% faster than those who didn't.

A significant opportunity lies in the rapidly decreasing token costs. True free market dynamics are at play: with increasing competition driving both quality improvements and cost reductions across the board — whether for foundational models (like Claude and OpenAI), coding agents (Bolt, Replit Agent), or AI-wrapper products.

Dr. Kadous advises against postponing AI integration due to current costs. What seems cost-prohibitive today could be feasible within months.

The strategic approach is to begin implementation now, allowing time to optimize integration while costs naturally decline.

6 Key Areas of Implementation

The first thing to understand is that every startup is unique — there’s no univeral answer of what the best areas of focus are. Identify your core processes, determine where efficiencies can be unlocked, and focus there.

Starting with that, you can then branch off to 6 general areas:

1. Productivity 

2. Customer Support

3. Accelerating Software Development

4. Rapid Prototyping

5. Multi-Linguality and Context Adaptation

6. Understanding & Working With Your Data

1. Business Operations and Productivity

The key is matching LLM capabilities to the particular tasks you deal with that are reliant on language.

For example:

• Marketing: Highly personalized marketing campaigns, tailored to specific individuals based on CRM data, could generate emails that feel personal and address unique concerns of different market segments.

  • automating content creation

• Contracts: You could upload a legal contract and ask it questions, a task that would otherwise require a legal assistant.

  • Examples of startups in the legal field:

    • https://www.harvey.ai

    • https://zuva.ai

2. Customer Support Enhancement

The impact of AI on customer support has been remarkable. Dr. Kadous cites Eric Brynjolfsson's research with a Fortune 500 company.

For example: you can create a voice agent, using your own voice, through play.ai, using your Customer Support docs.

Another interesting startup I came across is Pam, founded by Samee Khan, that acts as a receptionist for your card dealership, making sure you don’t miss calls but also booking appointments & answering questions.

The potential extends to in-product support, offering contextual help and natural language interfaces. As Dr. Kadous notes, "You may want to offer a natural language interface to your system for your customers. We also talked about in-context help - situations where you might offer a product and then help the user understand what they can do and cannot do in your UI in a very contextual way."

I’ve used this in Chatbase.co, which is the service I used to build the MuslimHub.co chatbot (on the bottom right of your screen!). Whenever I was having trouble, for ex. in training data, fine-tuning, I’d simply ask the in-product chatbot they’d created — no customer support tickets needed!

3. Development Acceleration

Recent advances in AI coding tools have been substantial, like;

1. Bolt.new

2. Cursor AI

3. Replit Agent

4. v0 by Vercel

5. Claude Artifacts

   1. check out some of the things people have made here

6. GitHub Copilot

While these tools don't replace programmers, they can:

1. Supercharge developer productivity, helping teams produce twice as much output or achieve higher quality work.

2. Reduce initial costs by potentially decreasing the immediate need for, a front-end developer, for example.
   

3. Eliminate tedious work, allowing developers to focus on more rewarding and mentally stimulating aspects that, drive technical innovation and product improvements

   1. I think this is the most important one by far!

4. Rapid Prototyping

One of the problems non-technical founders deal with is how to validate demand for an idea. A lot of founders unfortunately go down the route of spending a ton of money on an agency or hiring a developer, only to find out they don’t have product-market fit.

Dr. Kadous emphasizes the ease of prototyping with AI, using some of the tools listed in the section above.

This capability allows non-technical people to create MVPs without significant developer investment, enabling quick market validation before major resource commitment.

This is something that Abdellatif also emphasized in his advise on building startups!

5. Multi-Linguality and Context Adaptation

Multi-linguality is particularly crucial for Muslim businesses due to Islam's universal nature and diverse ethnic composition. AI can rapidly reduce market expansion costs by instantly translating website content and customer support materials.

Beyond simple translation, AI enables content adaptation to local contexts.

Dr. Kadous notes: "You can use AI even for markets that would previously have been too small to localize."

This extends to understanding local pain points and incorporating them into marketing materials, such as landing page calls-to-action. For businesses operating across multiple countries, AI can help process legislative changes and quality control reports, turning diverse information sources into actionable insights.

6. Leveraging Your Data

Perhaps the most overlooked opportunity is the potential to extract patterns from existing business data.

Dr. Kadous explains the process:

This capability extends to real-time data analysis. For example: ask your system what the top-selling product is in the last 24 hours, and bump it up to the top of the page or CTA on website.

The transformation extends from traditional dashboards to natural language interactions.

Dr. Kadous cited an interesting example of a company providing wireless network connectivity for supermarkets:

Ansari.Chat & Applications to Islamic Scholarship

Here I wanted to highlight a cool product that Dr. Kadous has built, and some ideas of how AI could be used in Islamic scholarship.

Ansari is an Islamic AI Assistant – a tool to help non-Muslims understand Islam and Muslims practice their faith more effectively.

Of course it goes without saying that it does not replace the role of a Shaykh. Taking advise from an AI system, or even most online sources is generally inadvisable, especially considering that AI’s often make stuff up. However, it is an improvement on online forums where Muslims regularly get erroneous information.

I’d recommend services like Seekers Guidance where common questions are answered by scholars.

But for quick, easily-verifiable questions, I think Ansari.chat is a useful service. It’s trained on the Qur’an, books of Hadith, and statements of scholars, including the Encyclopedia of Islamic Jurisprudence, a comprehensive 45-volume work. And it leverages RAG techniques to reduce hallucination, so you can be reasonably confident with simple questions. You obviously don’t want to jump into the deep end of the pool with inquiries on kalam or usul ul-fiqh.

Also, Muslims are using ChatGPT for questions like this anyway, which is definitely not going to be trained on the amount of data Ansari is trained on, nor with the care of avoiding biases that Dr. Kadous has taken great lengths to mitigate.

For example: I asked it about qabd and it provided a reasonable answer that many Muslims may not even be aware of.

Conclusion & Resources on AI in the Muslim Context

The integration of AI into business operations and Islamic scholarship represents a significant opportunity for innovation and efficiency. Whether in customer service, product development, or scholarly research, AI tools are becoming increasingly sophisticated and accessible. The key to success lies in thoughtful implementation, understanding local contexts, and leveraging existing data effectively.

Here are some resources that go deeper into AI in a Muslim context I can’t recommend this series of videos enough: Islam & AI Playlist by Masjid DarusSalam

For intro resources by Dr. Waleed Kadous;

• Introduction to AI by Dr. Waleed Kadous

• LLMs in Production by Dr. Waleed Kadous

My thanks goes out to Dr. Waleed, and the Muslim Biz Summit for making this information easy for us to digest.