Why We Build AI Businesses to Learn
The problem with AI advice that hasn’t been tested
There’s no shortage of AI consultancies willing to tell you how to adopt AI. Most of them have never shipped an AI product themselves. They’ve read the research, attended the conferences, and built slide decks. But they haven’t sat in the seat you’re sitting in — trying to make AI work inside a real business with real constraints.
That gap between theory and practice is where most AI adoption fails. Not because the technology doesn’t work, but because the advice doesn’t account for what actually happens when you try to implement it.
Our approach: build first, advise second
At Tenjin Factory, we take a different path. We rapidly prototype and launch AI-powered businesses — real products with real users and real revenue. Then we study what worked, what broke, and what surprised us. Those findings are what we bring to our clients.
Bangkok Inspect is one example. We built an AI-powered property inspection service for Bangkok’s expat community. The AI handles checklists, report generation, and certificate creation. But the lessons we learned building it — about prompt engineering in production, about managing AI outputs that need to be legally accurate, about integrating AI into workflows where human judgment still matters — those lessons apply far beyond property inspection.
CoffeeLog is another. An expertise-weighted coffee rating platform where a certified Q-grader’s review carries more algorithmic weight than a casual opinion. Building it taught us about AI-driven content moderation, weighted scoring systems, and how to design AI features that enhance rather than replace domain expertise.
What this means for our clients
When we advise a client on AI integration, we’re not guessing. We’re drawing on specific, recent experience:
- How long does it actually take to fine-tune a workflow around AI? We know, because we’ve done it across multiple products.
- What breaks when you move from prototype to production? We’ve hit those failures ourselves.
- Where should a human stay in the loop? We’ve learned the hard way where AI needs supervision and where it can run independently.
The compounding advantage
Every product we launch feeds back into our knowledge base. The more we build, the better our advice gets. It’s a compounding advantage that no amount of theoretical study can replicate.
This is why we exist as a company that both builds and teaches. The building isn’t separate from the consulting — it’s the foundation of it.
If you’re exploring AI adoption for your organization, we’d welcome a conversation about what we’ve learned. Get in touch.