---
title: "Mi-Memory: A Lifecycle Memory Framework for Personal AI"
source: "arXiv AI/CL/LG（论文）"
source_tier: "t2"
url: "https://arxiv.org/abs/2607.18975v1"
permalink: "https://feed.astralor.com/items/itm_2b9c49f565434fedd52b9e9b"
published: "2026-07-21T11:10:38.000Z"
score: 70
channel: "tech"
tags: ["Mi-Memory","Personal AI","Audit Contract","Memory Framework","Edge Deployment","Darwin Lab"]
---

# Mi-Memory: A Lifecycle Memory Framework for Personal AI

## AI commentary

> This paper presents a comprehensive lifecycle blueprint for personal AI memory, ideal for researchers in AI agents, edge computing, and privacy engineering. It reimagines memory from a 'conversation cache' into an auditable governance layer, addressing the lack of evidence chains and deployment awareness in current LLM memory systems. If you're designing multi-device AI assistants, Mi-Memory's architecture is worth deep study.

## At a glance

As personal AI expands from chat-only interaction to continuous services across phones, cars, homes, wearables, and tools, memory must evolve from a conversation cache to a durable, auditable, and governance-oriented substrate. Mi-Memory organizes around four roles: Structure, Expansion, Evolution, and Deployment, linked by an audit contract via typed evidence payloads, diagnostic traces, strategy artifacts, and gate/rollback records. In controlled Structure evaluations, MemStack achieves 93.59%, 57.24%, and 87.47% on LoCoMo, PersonaMem-V2, and LongMemEval respectively. The framework represents a step toward auditable, evidence-gated, and deployment-aware memory systems for personal AI.

## Key points

- Mi-Memory structures personal AI memory into four roles: Structure, Expansion, Evolution, and Deployment, with an audit contract ensuring traceability.
- Four components—MemStack, MemSense/MemFuse, D²ACCI/E²MEND, and LiteMem—realize each role respectively.
- In Structure evaluations, MemStack achieves leading results on three benchmarks (up to 93.59%); other tracks show preliminary or design-level evidence.
- The framework emphasizes evidence-gating and deployment awareness, adapting to latency, cost, privacy, and edge-cloud constraints.

## Full text

Personal AI is moving beyond chat-only interaction toward continuous services that span phones, cars, homes, wearables, cameras, and tools. In this setting, memory cannot remain a cache of prior conversations. It should serve as a continuity and governance substrate: preserving durable user state, grounding answers in multimodal and device evidence, supporting correction and forgetting, bounding policy evolution, and remaining deployable under latency, cost, privacy, and edge-cloud constraints. This technical report presents Mi-Memory, a lifecycle memory framework for Personal AI organized around four roles: Structure, Expansion, Evolution, and Deployment. A shared audit contract links these roles through four recurring artifact families: typed evidence payloads preserve source identity and provenance, diagnostic traces localize evidence loss across the serving pipeline, strategy artifacts make memory-policy changes explicit, and gate/rollback records bound accepted evolution. MiMemory instantiates the roles through MemStack, MemSense/MemFuse, D$^{2}$ACCI/E$^{2}$MEND, and LiteMem. In controlled-reference Structure evaluations, MemStack reaches 93.59%, 57.24%, and 87.47% on LoCoMo, PersonaMem-V2, and LongMemEval, respectively; other tracks report module-level, preliminary/internal, transfer-feasibility, or design-only evidence with explicit boundaries. MiMemory is a step toward auditable, evidence-gated, and deployment-aware memory systems for Personal AI. Project homepage: https://darwin-agent.github.io/Mi-Memory/ .

---

Verify against the original before citing; curated by Astral Feed: https://feed.astralor.com/items/itm_2b9c49f565434fedd52b9e9b