About · The Architect
The work is one idea, taken seriously.
I'm Jorge Guberte — AI systems architect and independent researcher in São Paulo, Brazil. For years, everything I've built has orbited a single conviction: AI systems should retain context, adapt over time, and remain useful beyond a single interaction.
Most of the industry optimizes the moment — better answers inside a single session, then total amnesia. I work on the other axis: what happens to an AI system across weeks, months, years. How it accumulates understanding. How it forgets well. How it stays itself.
The through-line
The work splits into a lab and a workshop. In the lab: LoomDB, a temporal memory architecture where context behaves like a living graph; EPCG, a formal framework for emergent, persistent agent personality; Ayvu-Talian, language modeling where the stakes of forgetting are cultural, not technical. In the workshop: Pixie, an embodied learning companion that proves the research in the hardest possible arena — a child's attention; and Multiverse, the open-source expression of how I think about state, history, and divergence.
None of these are side projects. They are one research program with multiple instruments.
The practice
Alongside the research, I'm co-founder and CTO of a GovTech company, where I architect AI systems for complex public-sector data — agent orchestration, retrieval pipelines, and long-running workflows that real institutions depend on. Production is my forcing function: it keeps the research honest, and the research keeps the production ahead of the curve.
Before all this, a long foundation in software engineering — platforms, APIs, infrastructure — which is why my AI work looks like systems architecture rather than prompt collections.
Where I operate
- Agent memory & temporal context architectures
- Context engineering & retrieval systems
- Cognitive architectures & agent personality
- Agent orchestration for long-running applications
- Embodied AI & human-AI interaction
- Accessibility-first, neurodivergent-first interaction design
I like systems that feel alive: things that remember, decay, branch, adapt, and become more useful with time. If you're building something that needs to last longer than a session, we should talk.