April 2026 marks one year since BlueForge Studio shipped its first product. A lot has changed since then.
We started with a simple hypothesis: that interconnected AI-powered products, built shared infrastructure, and cross-referring users would outperform a traditional single-product startup. One year in, the data is more nuanced than we expected — but the overall thesis is holding.
**What Worked**
**The ecosystem approach**: Users from QuestLang Thai who discover GameVault convert at 3x the rate of cold traffic. The ecosystem flywheel is real.
**Vertical SaaS in underserved markets**: GarageOS reached $12K MRR in 4 months with zero paid acquisition. When the market fit is right, go-to-market is almost instant.
**Shared infrastructure**: Running 40+ products on Supabase, Resend, and our internal tooling stack means our per-product overhead is 60% lower than if each product had its own setup.
**What Didn't Work (Yet)**
**Adventure Craft**: Browser-based multiplayer voxels at scale is technically harder than we anticipated. We've had to cut scope significantly. A v1 launch is still 3-4 months away.
**Creator Gold membership**: Our first attempt at a bundled subscription product didn't resonate. Creators want per-product purchase, not ecosystem subscriptions. We're restructuring to individual product sales with ecosystem discounts.
**Geographic expansion timing**: We tried to launch NordicQuest while still iterating on ThaiQuest. Splitting attention slowed both. Lesson learned: nail product-market fit in one market before expanding geographically.
**What We're Still Figuring Out**
**The right org structure**: At 41 products, we're hitting the limits of what a small central team can effectively oversee. Do we spin products into separate companies? Hire product-specific teams? The answer is probably somewhere between both.
**Pricing consistency**: Some of our products are free, some are $9/month, some are $299/month. We don't have a coherent pricing philosophy yet — just a lot of one-off decisions.
**The AI integration question**: How much AI is too much? We're using GPT-4 for content generation, Claude for code review in Forge Git, Gemini for image analysis in GarageOS, and rolling out custom models for QuestLang speech recognition. But the AI costs are real, and the pricing models aren't stable.
**What's Next**
Year two will be about focus. We're not going to 100 products — we're going to make the 20 most promising products genuinely great. We're building the membership tiers (Silver and Gold) to give users a reason to go deeper into the ecosystem. And we're preparing for our first institutional raise.
To everyone who's been on this journey with us: thank you. The next year is going to be interesting.