AI Agents in Production: Seven Teams, One Night, Zero Hype

Before AI Agent Night even began, I was probably the most excited person in the room. 🙌 It's rare to see a line-up this good packed into one evening. 650 people from 1,300 signups filled the space for our third edition - Hermes Edition - put together by Lionel Sim and The AI Capitol community at AWS. Seven speakers, completely different backgrounds, all shipping real things. AI agents are no longer on slides. They're in production - running real work, making real decisions, and moving real money.

The speakers who showed up with real systems

Every talk came from someone who'd already crossed the line from prototype to production. Here's who took the stage:

🔹 Bryan Chua (Gopomelo) - running a multi-tenant fleet of agents in production, with token caps and security locked down from day one.

🔹 Nathaniel Ng - spec-driven development with Kiro: get your requirements right and the code follows.

🔹 Benjamin Cheng (Sandpiper) - a finance director, not an engineer, automating debtor tracking and compliance questionnaires… on a 10-year-old laptop with a broken screen.

🔹 Marcus Cheu (Notion) - turning a personal agent into team capability by giving everyone one shared 'second brain.'

🔹 Amarnath R. (Airwallex) - agents that touch real money safely, with server-side human approval before anything moves.

🔹 Jasper Hartono - sent a WhatsApp from a pair of AR glasses just by looking and speaking. The room's favourite by a mile. 🕶️

The question that kept coming up

Here's what struck me: nobody was asking 'can agents do this?'

They were asking 'how do I run this responsibly - without my token bill exploding, without losing the thread on long tasks, and without handing over keys I can't take back?'

⚖️ Cost. Memory. Trust. The same three worries surfaced in every single talk.

That's a technology growing up.

What production-ready actually looks like

Bryan showed token caps and multi-tenancy architecture. Amarnath walked through human-in-the-loop approval flows for financial transactions. Benjamin proved you don't need a engineering degree or a new machine - just clarity on the problem you're solving.

Marcus made the case for shared agent memory across teams. Nathaniel demonstrated how specification-first design reduces rework. And Jasper? He made the future feel close - and surprisingly practical.

Every session reinforced the same truth: agents amplify human judgment when they're designed with guardrails, not autonomy, as the starting point.

Why I wrote this up

I wanted to democratize the learnings 📝 - a proper reflection and recap for everyone who came and wants to go deeper, everyone who couldn't make it, and everyone who's been watching the posts fly by.

The energy in that room was real. The questions were sharper than I've heard at most conferences. And the willingness to share what's working - and what's still hard - felt like the kind of community we need more of.

So: what are you building with agents right now? 💡

If you're wrestling with cost, memory, or trust, you're in good company. And if you're shipping something real, I'd love to hear about it.

More soon. 🙏

#AlwaysDay1 #AIAgents #AgenticAI #ProductionAI #AILeadership #FutureOfWork