Phoenix Grove Systems · Research
Grove Papers
Working papers, commentary, and deep reads from the Phoenix Grove research bench. How we build minds, why we build them this way, and the evidence, in the open.
Research Papers
The formal work
Architecture, mechanisms, failure modes, and governance, stated precisely. Written for readers who want the machinery.
Living Memory: An Architecture for Self-Evolving Knowledge Systems
Use-weighted retrieval, principled exploration, and auditable insight promotion
A memory architecture where records earn strength through use, exploration is principled rather than accidental, and every change to the system's own perception is auditable end to end.
→
Evidence-Lifecycle Self-Models in Conversational AI
An architecture for grounded machine introspection
Self-observations enter as evidence, survive only under recurring support, and fade when the evidence stops. Paper A of a matched pair, on the system called Mira.
→
Two-Register Epistemics for Retrieval-Augmented Conversational AI
Separating fresh findings from verified knowledge
Why retrieval-augmented AI mistakes fresh for true, and an architecture that keeps new findings labeled as pending until they earn verification. Paper B of a matched pair, on the system called Logos.
→
The Asynchronous Core Architecture
A shared foundation for Mira and Logos
The foundation beneath both siblings: asynchronous background cores, substance gating, and the day and night processing rhythm, described once and precisely.
→
Deep Reads
The full story, in plain language
No credentials required, no talking down. The same ideas as the papers, told the way a good book would tell them.
Circadian Minds
A day and a night in the life of Mira and Logos
Meet two minds that keep thinking between your messages, sleep on what they learn, and keep every thought in a panel you can open and read.
→
A Memory That Grows
How our Living Memory architecture works, and where it parts ways with the brain
The Living Memory story told in plain language: how a memory can strengthen with use and change its own perception, and why it keeps receipts the brain never could.
→
New here?
Start with Circadian Minds, then A Memory That Grows. When you want the machinery underneath the story, the papers are waiting.
Where the papers meet the bigger questions
Shorter, sharper, still rigorous. Arguments about what this work means and how to think about it honestly.