The AI listening to your pastor on Sunday morning was built by a Nigerian in Ibadan.
Dara Sobaloju built PewBeam AI in public, starting from a birthday reflection. What he built is becoming the communication layer for how churches operate, and that is just the beginning.
On August 21, 2025, Dara Sobaloju posted a tweet that was not supposed to be a product announcement. He was not ready to build. He just wanted to see if the idea was worth building.
“I want to build a Bible presentation AI agent for church use,” he wrote. “Imagine Bible verses coming up on screen as the pastor preaches just based on what he’s talking about or his paraphrases and quotes.”
The replies filled up before the afternoon was over. Pastors. Developers. Churchgoers. People who had experienced the exact problem he was describing every single Sunday, projection teams scrambling to keep up with a preacher who spontaneously quoted scripture, the congregation losing focus mid-flow, the moment broken. The problem was real. The demand was immediate.
“When I saw the fact that a lot of people already see this as a legitimate problem,” Dara told me, “I got started on it.”
Six months later, PewBeam was live, a finished desktop application available on Windows and macOS that listens to a live sermon and automatically displays the exact Bible verses being referenced in real time. No preloaded slides. No manual searching.
Just AI listening to the pastor and matching their words, including paraphrases and spontaneous quotes, to the right scripture within 80 milliseconds.
Today, over 7,100 churches across more than 20 countries use PewBeam. About 70% of Nigeria’s biggest churches are among them. And the paying customer split between diaspora and Nigerian churches sits at roughly 60/40, which means the Nigerian churches that everyone assumed would never pay for a subscription are actually showing up.
“It was also a concern for us,” Dara said about the subscription question. “But surprisingly, it’s been amazing.”
What PewBeam actually does, and why it works where others didn’t
PewBeam’s design focuses on three things: real-time speech-to-text processing with minimal delay, an offline-first architecture that works without internet connectivity, and contextual accuracy, meaning the system understands the pastor’s intent, not just the keywords they use.
That offline-first approach is not a technical footnote. It is the entire value proposition in a country where internet access during church services cannot be assumed. “Anything that can work in Nigeria can work outside Nigeria,” Dara said. “It’s a harsh environment.”
The speed has also been pushed to a point that sounds almost absurd. Early versions processed in around 500 milliseconds. The team brought that down to between 2 and 100 milliseconds, fast enough that the verse appears on screen almost before the congregation consciously registers the pastor has quoted it.
One evening during our conversation, Dara mentioned a customer in the UK who was using PewBeam live during a service and needed a feature that did not yet exist. The team built it and shipped it in under 30 minutes. The customer used it the same night.
That iteration speed, from customer request to live deployment in under half an hour, is not something most startups can claim, let alone sustain.
Their retention rate reflects it. Close to 80% of paying customers stay. For a product that is still iterating, in a market that everyone assumed would resist paying for church software, that number is striking.
The bigger vision, and why it starts with the church
Dara is building something that sounds specific but is actually much larger.
For most of history, the tools churches use were built for the secular world and repurposed for worship. Projectors. Livestreaming software. Presentation tools. All designed elsewhere and adapted for church use. PewBeam is trying to reverse that direction, build something specifically for the church that eventually finds application in the wider world.
“We are trying to flip the switch,” he told me. “We’re trying to build something for the church that will eventually find a use case in the secular world.”
The newest product from the team is PewNote, launched at their Lumen event and not yet publicly available, which gives church members access to sermon transcripts they can interact with, ask questions of, and engage with between services.
The goal is to extend the communication layer from the moment of worship into the days that follow.
In five years, Dara wants PewBeam in at least 120 countries. He also wants to build a proprietary AI model, one trained specifically on the nuances and context of scripture, that faith-based developers can build on top of, the way developers build on OpenAI’s models today.
“Before the pastor thinks about the scripture,” he said, only half-joking, “the scripture is up.”
The question nobody in Nigerian tech is asking
There is a broader trend here that deserves attention. PewBeam is not alone.
Other Nigerian developers are building in the same space, Tolulope Adeniyi is building Spetra, a similar real-time verse detection tool, while Olanrewaju Taiwo built Meno, a Bible app powered by Google’s Gemini model that explains scripture in context.
Deeper Life Ministry launched Ask Kumuyi, an AI chatbot for sermon questions.
Nigeria has some of the largest church congregations in the world. Hundreds of millions of people across Africa gather in physical worship spaces every week.
The communication infrastructure for those gatherings has historically been underbuilt. AI is changing what is possible, and Nigerian developers are, for once, not late to it.
Dara put it in terms that are difficult to argue with. He remembered when some churches called television the devil’s box. Today, YouTube carries more live church services than any physical building could ever hold.
“This time around,” he said, “the church should not come late to the reality of AI.”
With over 7,100 churches already using PewBeam, it might already be ahead.