The LLM progress of early 2026 changes the direction of Blue-Lila.
That includes Gen-code, X-Go, Kit-Cla, My-Soft, and My-Game.
X-Go, Kit-Cla, and Gen-code were built to help individuals make personal apps and community game tools with far less time and specialised knowledge.
Because a small app once meant 50–100 hours; a polished one, 200–500 — and required a skilled developer.
Mostly robot-generated code, configurable through a GUI and sometimes augmented with custom or LLM-generated code.
The large LLM companies’ all-in-one model: a general-purpose conversational engine that writes, explains, and adapts code on the fly.
Gen-code still wins on determinism and consistency; LLMs now win decisively on flexibility.
Many are no longer needed in their current form.
The projects continue, now with LLM-assisted development.
Still relevant, with LLM integration ahead.
The LLM becomes the real user; the human becomes the pilot.
LLMs are changing both code and asset production.
The lightweight, moddable X-Go app remains the long-term aim.
New work around LLM-based workflows, alongside innovative projects such as Caravel, carries Gen-code’s lessons into the LLM age.
This is the direction, not a finished transition: time has been short since early 2026, so most of the major changes are still ahead.
I liked writing code by hand a lot. I liked building these tools. I spent hundreds of hours, sometimes thousands over a decade, on some of them. Now they are obsolete or need to change. Letting go is bittersweet. But resisting the shift would not help.
There are things I dislike about LLMs. Plenty still needs proving.
Ironically, I was an early LLM user. Before ChatGPT took off, Gen-code already used early models.
I never expected them to become this capable.
But I’m excited too, for many different reasons.
It achieves part of the original goal. And even more. Just in a very different way.