In 2015, we started Carro as a used car marketplace in Singapore. By 2025, we'd grown to over 4,000 people across six countries, processing more than 10,000 cars a month. Along the way, we built over a dozen software systems from scratch - CRM, fleet management, inspection apps, dispatch, inventory, credit management, leasing, auction systems, workshop management, and more.
We didn't plan to build all of that. We tried the off-the-shelf tools first. Salesforce, HubSpot, Zendesk, Freshdesk, Intercom, MessageBird - we went through all of them. Every time, the same problem. The tool didn't understand how our industry actually worked, so we'd spend months configuring widgets and plugins, fighting to bend it into something usable. Eventually we stopped trying and built everything custom.
It took a decade. Hundreds of engineers. Millions of dollars. And honestly, we're proud of what we built. But I kept thinking - does it really need to be this hard?
What Changed
I've been writing software since 2003, and nothing I've seen in those years comes close to the shift that's happened since 2023. AI went from autocompleting lines of code to understanding entire codebases and executing full objectives across a project. A team of four recently shipped a hire purchase management platform in about four months - loan origination, payment tracking, collections workflows, the works. Two years ago, that same project would've needed fifteen people and a year.
The systems we spent a decade building at Carro could now reasonably be built in a fraction of the time and resource. Not because AI writes perfect code - it doesn't, and we've got the scars to prove it. But because it eliminates the boring parts. Boilerplate, test scaffolding, documentation, migration scripts. Strip that away and suddenly a small team of experienced people can focus entirely on the hard problems - architecture, business logic, the edge cases that actually matter.
What We Believe
So we started First To Fly with a question: if the cost and time to build software has dropped this dramatically, what becomes possible?
Here's what we think. The bottleneck isn't building anymore - it's knowing what to build. The people with the deepest understanding of an industry - the operators who've spent decades learning the edge cases, the workarounds, the unwritten rules - they're sitting on the most valuable input for software. They just haven't had a team that can turn that knowledge into a product without a million-dollar budget and a two-year timeline.
That's the gap we're filling. We co-build vertical software with industry partners. Not as an agency that takes a brief and delivers a build. As a partner that sits alongside domain experts, shares the risk, and stays in the codebase long after launch. We bring the systems experience - a decade of building CRM, scheduling, inventory, fulfilment at Carro. They bring the industry knowledge we can't Google.
What This Blog Is For
This isn't corporate content marketing. We're operators who write about what we're actually doing, and we want this to be genuinely useful.
Here's what you'll find here:
-
What we're learning - Real observations from building AI-native software. What works, what doesn't, what surprised us. We've had AI put the same logic in three different files and only update one of them when asked to make a change. We'll share the honest version, not the polished one.
-
Industry perspectives - We talk to businesses across Southeast Asia every week. Logistics companies, clinics, dealerships, food distributors. Almost every one of them is running critical operations on spreadsheets and WhatsApp. We'll write about the patterns we're seeing and where we think industry software is heading.
-
Guides and frameworks - If you're new to AI or trying to figure out how to actually use it in your business, we'll share practical guides based on what's worked for us. Not theory - real workflows, real tools, real trade-offs. We think AI needs to be deployed in layers, not as a one-size-fits-all solution, and we'll break down how we think about that.
-
Building in public - Product updates, architecture decisions, things we got wrong. We believe the best way to build trust is to show the work.
We're Just Getting Started
First To Fly is early. We're a small team, we're building fast, and we're learning constantly. Some of what we write here will probably look naive in hindsight. That's fine. We'd rather share what we're thinking now and correct course later than wait until everything is polished.
If you're an industry expert who's been thinking about building software for your domain, or if you're just curious about where AI-native development is heading, we'd love to hear from you.
Let's build.
