A PM on a client team in Japan put it to me directly — and honestly, I’ve heard versions of it for two years now: “If Claude Code can write most of this, why are we still paying an offshore team?”
Honest answer: half of that is right.
I run a development team in Ho Chi Minh City. Internally, we let everyone use Claude Code freely — there’s no reason not to. It made us faster. For a prototype, a landing page, an admin screen, Claude Code or Cursor can get something working overnight now. Individual productivity went up, clearly. So “offshore because the labor is cheap” is mostly over. I’ll admit that openly.
But here’s the thing. “An individual is fast” and “a team ships the same quality no matter who does the work” are two completely different things. If you only look at the convenience, that gap disappears from view.
Convenience scales to everyone. Quality doesn’t.
This was the part that actually scared me.
Before, someone who didn’t understand the work simply couldn’t produce. So you knew right away. With AI, a person can produce a polished-looking result without understanding it. On a distributed team — where you can’t lean over and look at a screen — that’s ten times more dangerous. From the output alone, I often can’t tell who actually understood the work and who just shipped what the AI handed back.
The code compiles, it runs, it looks fine. Then you ask, “why did you write it this way?” — and there’s no answer. I’ve seen this more than once on my own team.
The younger the engineer, the easier it is to feel “done” because the AI handed them an answer. The scary part is there’s no bad intent — the tool is good enough that they can’t see what they don’t know.
Convenience reaches everyone the moment you hand out the tool. Quality — the ability to catch the AI’s 60-point output and lift it to a production-ready 100 — doesn’t spread on its own. If anything, the better the tools get, the harder it becomes to see who understands and who doesn’t. The problem didn’t disappear. It just went invisible.
So what do you actually distribute? Only the system.
Not convenience. Not effort. A system. “Win with systems” sounds hollow as a slogan, so let me tell you what we actually did.
First, we made the good ways of using AI a team standard, not a personal trick. When someone finds a sharp way to use Claude Code, it becomes everyone’s procedure — the exact opposite of using it quietly on a personal account.
Second, we decided who passes the AI’s output last. We have something we call Self-QC: you build quality into your own step and don’t pass defects downstream. The AI’s 60 always goes through a person who understands it. “The AI made it, so it’s probably fine” is banned structurally, not by goodwill.
And third, we kill the hero dependency. This was my belief long before AI — turn “nothing works without that one person” into “anyone can do it the same way.” The irony is that AI, by erasing the scarcity of individual skill, pushed the whole game onto the system side. You can’t win on heroes anymore. So you win on systems.
Adopting AI isn’t buying a tool. It’s re-wiring the team.
Hand the same Claude Code to two teams and they split. One where it stays an individual weapon, and one where it becomes the same quality no matter who does the work. Whether you can build the second kind is, the way I see it, the value of offshore development from here on.
One honest thing, though. Building this system from scratch takes time. We’ve been at it for years, failing along the way. So there are really two paths: build it into your own team, or work with a team that already has it. Neither is “the” right answer — pick based on where you are.
And if you’re wary of offshore quality, that’s fair — honestly, you should be. It’s exactly why we put a QA checkpoint before anything touches your production environment, and why we don’t hand off and disappear. Senior engineers in Ho Chi Minh City, a PM who speaks both the technical side and the business side, at roughly 30% of an equivalent Tokyo build. The cost was never the point. The system is.
So where does AI actually create an edge?
What I keep coming back to is this: the convenience is the part anyone can hand out now. So the edge stopped living there. The edge is whether you can turn it into a team system — AI you use alone and AI a team uses with reproducibility are two different animals.
This is the same thing I’ve been saying since before AI showed up. Speed alone doesn’t win. Whether you get the same result no matter who does the work — that’s it. We just call it Self-QC and keep doing it, quietly. AI didn’t change that. It only made that quiet, unglamorous work worth several times more.
FAQ
With Claude Code, can’t we just build in-house instead of using an offshore team?
You can build prototypes in-house — individual speed belongs to everyone now. But team-level, reproducible quality doesn’t come from handing out the tool. It comes from a system: standardizing how AI is used, a human QA pass on every AI output, and removing single-person dependency. If you already have that system, build in-house. If you don’t, a team that already has it gets you there faster. See how we build development teams.
Does AI improve or hurt offshore development quality?
Both, depending on the system around it. AI lets people who don’t fully understand the work ship polished-looking output, so the gap between “looks done” and “actually understands” hides — and it hides worst on distributed teams. A structural QA pass — we call it Self-QC, where someone who understands signs off on AI output — is what keeps quality real instead of just plausible.
What is an “AI-native offshore team”?
A team built on the assumption that AI writes the code: AI use is standardized across the team, every AI output is quality-assured by a human, and hero-dependency is removed. What it sells isn’t speed or convenience — it’s reproducibility. You can talk through what that looks like for your team in a 30-minute session.
Is your team using AI as a personal weapon — or as a system?
If this resonates, I’m happy to talk through what it would look like for your team — building an offshore setup that ships reproducible quality, not just convenience.
Book 30 minutes →Related reading
Shogo Harada原田 祥吾
CEO · Linnoedge Inc. · LinkedIn↗
Operating IT offshore development and overseas expansion support businesses across two bases: Tokyo and Vietnam. A leader who believes in “Systems over Spirit,” structuring cross-border businesses that often tend to be opaque. Committed to providing “reproducible quality” to organizations and clients rather than relying solely on individual skills.