OUR MISSION
From Japan,we will end the multi-layer subcontracting structure.
We want to eliminate companies that lose out simply because they don't know better. We want to build a world where value flows straight through — to engineers and clients alike.
Agile Technology Inc.
CEO Kazuki Nakai
VISION
In-house development,for every company.
It's fine to order from a vendor. What matters is understanding what's inside, never just going along with it, and keeping the reins in your own hands. "A state where you judge for yourselves, and put it to real use" — that's what we call “in-house development.”
About the CEO
Kazuki NakaiKazuki Nakai
CEO. After a stint in sales at Recruit and a failed startup, joined Microsoft Japan and NTT Data. Has worked in development on every side — client, vendor, and in-house — and has watched the same scene repeat for 20 years. Went independent in 2019, incorporated in 2022. Supported agile development standardization at Red Hat. Now builds and runs 10+ products, one person + AI.
20 years
In the IT industry
10+
Public OSS projects
100+
Largest team led
Every layer
From sales to implementation
What I've seen
Japanese SIerscan only handle “contracted work.”
Contract work = a construction-industry method
Delivery is the goal.
Just like building a building. Completion and handover are the “goal.” After that, it's not their problem.
What software development should be
The start is after delivery.
Improvement only begins once something is actually used. Delivery isn't the goal — it's the starting point.
Here lies a fundamental conflict of interest. The client wants it “better,” the vendor wants to “deliver fast and be done.”
When “contract work” fixes the price and deadline,waterfall is the only option left.
Waterfall has hit its limit.
Decide everything up front, then flow one-way from upstream to downstream. No going back. — But reality doesn't work that way.
The future can't be predicted
Could you have predicted COVID five years ago, or the iPhone twenty years ago? No one can write a perfect spec up front.
You don't know until you use it
"I want this feature," "this button is hard to tap, move it." Without feedback, you can't build something good.
Bugs and defects get buried
Even when a design flaw is noticed mid-implementation, it goes unreported to protect the deadline and the payment. The Seven Pay incident is a textbook example.
Are you really okay handing your core technology entirely to another company?
Where the money goes
And on top of that, half the moneynever reaches the people doing the work.
Every time contracted work is handed down through another layer of subcontractors, a middleman margin gets skimmed off. By the time it reaches the team actually building it, there's very little left.
Client
¥10M/mo
Prime contractor
¥8M
2nd-tier sub
¥6.4M
3rd-tier sub
¥5.1M
Build team (5 people)
¥4.1M
Each layer skims off about 20%. After just 4 layers, over half the order amount is gone to margins, and only about 40% reaches the team actually building it.
The real solution
The solution isagile development.
Contract work fixes the deadline and the price. So it can't accept change, and becomes waterfall. Agile is the opposite.
Waterfall
Decide everything up front. Can't change.
Delivery is the goal.
Agile
Build small, use it, fix it.
Assume change, improve through feedback.
But agile is“hard.”
Do you have great people?
A top-tier PO, scrum master, engineers, and an understanding of agile itself. Assembling all of that isn't easy.
Can you delegate authority?
A self-driving team needs delegated authority. But in a Japanese organization, the decision to “trust the front line” rarely comes down.
Outsourcing can't cover it
Even starting with a lab-style contract or SES, you hit walls of talent and contracts. Handing it all off never produces something good in the end.
And that “in-house development” wasexpensive.
If you can't rent it all at once, you must own it yourself. In practice, both cost and dependency increased together at project start.
In-house owner who can make product decisions
One response map for phone, email, forms, and SNS
Internal owners for valuable, personalized outreach
Current baseline is1 person + AI as the practical starting point, with escalation rules defined.
But AI changed it
That ¥10M —the AI Front Office model changed it.
AI Front Office means handling both Contact (incoming) and Reach (outgoing) in one product logic, where outgoing is limited to valuable, personalized outreach. Value-less sales are filtered out, refusal text and reCAPTCHA are respected, and AI drafts answerable responses first, with humans reviewing policy-sensitive and human-only conversations.
BEFORE
7-8 people
¥10M / mo
AFTER
1 person + AI
From a few hundred thousand yen / mo (with human verification)
What AI lacks is your company’s context and clear refusal policy.AI handles response-first workflows, while humans own contextual judgment and final decision points.
My stance
Not dependence,but equals.
I aim for a model where role boundaries areAI-first on routine, human-first on policy so we can keep meaning in human dialogue.
Work with me, andthree things change.
Less low-value outbound noise
Reach and Contact are not contradictions. Reach is intentional, high-value proposal contact only. Contact is for inbound requests. This avoids one-directional spam and keeps dialogue useful.
Faster, safer replies
AI now produces first-pass answers quickly, and humans review final responses for sensitive, legal, or high-context cases. Meaningful conversation time is returned to your team.
Channel unification by phase
Current: one intake path for phone, email, and forms. Release-gated: SNS/owned media routing in the same model. Vision: outbound cadence and AI completion expand only where governance is verified.
Growing the number of “companies that can judge for themselves,” one at a time.
That's my mission.
Can your external touchpoints stay helpful and safe at the same time? We define what is current, what is release-gated, and what is still vision, starting from phone, email, form, and SNS.
