Agrim Singh is Making AI Something Anyone Can Build
Agrim Singh discusses the community component inherent to navigating AI building.
The world of AI is a lot more than just ChatGPT and Gemini. Its applications are broader, and way beyond just what we see on our screens. It works quietly behind the scenes, doing everything from assisting e-commerce platforms with the adjustment of prices in real time to directing the robots moving packages around warehouses. The more you look, the more you realise how deeply it is already entrenched in the systems that undergird our daily lives.
And Agrim Singh is someone who has been watching all that take shape right here in Singapore. Singh is the co-founder of 65labs, a community operation focused on bringing the best of AI to the builder ecosystem. Since it organised its first meet-up in 2025, it has grown into a regional community of around 10,000 people, bringing together everyone from experienced engineers to school teachers, HR professionals and complete beginners.
In May this year, the community organised one of its biggest events. The flagship US conference AI Engineer held its first conference in Singapore, bringing more than 1,000 AI builders and practitioners to The Capitol Kempinski for two days. Headline sponsors included some of the biggest names in the industry, including OpenAI, Google DeepMind, Cursor and Z.ai.
Singapore’s Minister for Foreign Affairs, Dr Vivian Balakrishnan, opened the conference by demonstrating a personal AI agent he had assembled himself on a Raspberry Pi 5. This agent worked as a second brain, processing his messages and voice notes, storing information, researching topics, helping with briefings and even drafting speeches. Dr Balakrishnan hadn’t developed a new AI model from scratch; he had merely taken existing open-source tools and pieced them together to meet his own needs.
And perhaps that is the most interesting thing about the current AI moment: that one doesn’t necessarily need to be an AI engineer to build with AI. It is about understanding what the tools can do, knowing which pieces to put together to achieve a solution and having the curiosity to see what happens when you do.
In many ways, Dr Balakrishnan set the tone for what the conference, and the wider AI builder movement, was about. This wasn’t your typical industry conference, where executives sit through panels discussing what the future might look like. There were no lengthy fireside chats. Instead, practitioners and builders took to the stage for 10 to 15 minutes at a time to show what they were actually working on.
The person behind ChatGPT and Codex was there. So was Cursor’s former head of design. One could simply bump into them in the crowd and start a conversation. No special access needed, just guts and curiosity. The event was deliberately designed to be practical and information-dense, with less speculation about the future of AI, and more of what people are already building.
Singh first encountered the local AI community back in 2013, when it felt very different from what you see today. Hackathons were regular fixtures. Start-up open houses were commonplace. “If you were young and curious, there was a lot to learn,” he says.
Over time, though, Singh felt some of that energy begin to disappear. The scene became increasingly focused on who was being seen, rather than what people were actually building. And that was what prompted him to create 65labs.
The idea was simple: to bring some of that old energy back. For him it wasn’t about forging business networks or organising corporate events, but creating a place where people could meet, build, experiment and learn from one another.
“We wanted to gather people for no-bullshit, no-frills set-ups. No yapping, thought-leadership kind of stuff,” Singh says. “It’s for people who are building using the tools, getting their hands dirty, and wanting to share [their experiences] with other people who are genuinely interested.”
That philosophy has been there since 65labs’ first event in July 2025 held at the office of ride-hailing platform Grab. “We had 500 people coming through the door. It was insane, given that it’s Singapore,” Singh says.
The programme that night was deliberately simple. Grab shared how it was using AI at an enterprise level. A start-up founder showed how he was building his company with AI. And a two-weeks old builder demonstrated a simple receipt generator. It was a place for people to learn from one another, and more importantly, you didn’t have to arrive knowing anything at all.
In less than two years, 65labs has grown to include around 10,000 people across the region, with two meet-ups a month.
Perhaps the biggest surprise is the people who turn up. You might expect the room to be dominated by the stereotypical tech crowd: engineers, developers and start-up founders. And yes, they might be there but along with them you’ll find a whole host of people from across generations and professions, from 13-year-old students to 63-year-old retirees. There are teachers, HR professionals and complete beginners alongside experienced engineers. Many come not knowing how to code — they’re just simply curious.
Over time, Singh has watched the mindset of the community change, too. In the beginning, people might have shown up for the “free AI tools credits and pizza”. Now, members contribute talks, organise niche meet-ups and even start companies together. What makes him proudest is that the community no longer depends entirely on the people who started it. “This scene shouldn’t start and end with us. If I can’t continue doing this tomorrow, is it self-sustaining?” he says. “I think I can now feel good about it being self-sustaining.”
Still, even if you talk to ChatGPT or Gemini every day, the wider world of AI can feel intimidating. There is an endless stream of new terminology, new models and new acronyms. Half the battle, Singh believes, is simply getting past the jargon.
He gives the example of hiring an intern. You give them access to past reports, show them what a good report looks like and provide some context about what goes into making one. It sounds obvious. “That’s what you do when you’re making agents as well,” he explains. “But in the AI world, this is called harness engineering.”
“If I told you what harness engineering is, you’d just be like, ‘What?’ But conceptually, this is very obvious,” Singh adds.
And perhaps that is the best way to think about AI right now. Behind the intimidating terminology and breathless predictions, much of it is simply people figuring out better ways to get things done. You don’t have to understand everything. You just have to be curious enough to learn and start building.
