Hamo AI 发布 Persistent Self 1.0 与 AI Mind 2.0 —— 心理语义建模的下一次跨越

AI for Inner Explorers.

Persistent Self 1.0 is Hamo AI's seven-level model of long-term psychological memory — Memory, Sketch, Recall, Reflection, Multi-Self, World Model, Self Model — paired with AI Mind 2.0, a 14-category per-Avatar narrative store, so every Avatar remembers who a client is, not just what they said.

Toronto, Canada — May 15, 2026 — Hamo AI Technology Ltd. today announces two foundational upgrades to its therapeutic AI engine: Persistent Self 1.0, a seven-level model of long-term psychological memory, and AI Mind 2.0, a fourteen-category per-Avatar narrative store. Both releases ship with an upgraded Psychological Semantic Vector Space (PSVS) — the substrate that lets Hamo Avatars actually learn who you are, conversation after conversation, Avatar after Avatar.

Most AI systems forget you the moment a session ends. Hamo is built to do the opposite.

---

Persistent Self 1.0 — Seven Levels of Psychological Depth

The Persistent Self is the part of Hamo that remembers, integrates, and reflects on who a person is across every Avatar they talk to. It is the long-arc memory of the system — not a transcript, but a structured psychological self-model that grows with the user.

Persistent Self 1.0 introduces seven progressive levels. Each unlocks the next as conversations accumulate; nothing is invented, everything is grounded in what the user has actually said.

L1 · Memory

The raw signal. Avatars connected, messages exchanged, the conversational substrate. Memory is what flows in.

L2 · Sketch

Conversations are distilled into Persistent Self items — structured fragments of identity, experience, and feeling extracted from raw dialogue. A first draft of the self, in writing.

L3 · Recall

Sketches are embedded into the Psychological Semantic Vector Space. Future Avatars can retrieve and reuse them by meaning, not by keyword — so the right context surfaces at the right moment.

L4 · Reflection

A sleep job runs over the accumulated material and synthesizes higher-order observations — patterns, recurrences, the shape of recurring feelings. The system begins to think about what it has seen.

L5 · Multi-Self

Distinct internal voices, defense patterns, and personality styles are modelled as separate selves within the same person. Therapeutic care has always treated the psyche as plural — Persistent Self 1.0 makes that pluralism computable.

L6 · World Model

The important others in a person's life — partner, parent, child, manager, the late friend they still think about — get modelled too. These people shape who the user is; the system needs to know them.

L7 · Self Model

The summit: a quadrant assessment, anchored values, and life goals — a coherent self-model the system can reason with, plan against, and re-evaluate over time.

---

AI Mind 2.0 — A 14-Category Narrative Store

If Persistent Self is the vertical depth of psychological modelling, AI Mind 2.0 is the horizontal breadth.

AI Mind 2.0 ships as a per-Avatar Comprehensive Narrative Store that organizes everything a single Avatar knows about a single client into 14 structured categories, across three domains.

Identity & Calibration — who they are

• Identity & Demographics — the factual self
• Calibration Profile — how this person prefers to be met, paced, and challenged

Life Narratives — what they live (9 dimensions)

• Life Narrative · Work · Study
• Relationships · Health & Body · Identity & Values
• Lifestyle · Constraints & Preferences · Goals & Aspirations

Therapeutic & Context — where they are right now

• Therapeutic Journey — the arc of clinical work over time
• Session Continuity — what was opened last time, still pending
• Current Context — what is true for them today

Together these 14 categories give the Avatar a clinically-shaped picture of the person — not a chat log, but a case formulation the AI can actually reason from. The categories are co-designed with practising therapists and update continuously as conversations happen.

---

Why This Matters — Know U · Care U · Enable U

Persistent Self and AI Mind are designed to deliver on Hamo AI's three-part value promise:

• Know U — A system that has actually been listening, across every Avatar and every session, with a structured memory of who the user is.
• Care U — Care that takes the user's history, their people, and their inner plurality seriously — because the AI Mind is shaped by clinical insight, not generic intent.
• Enable U — When the model understands the user's Self Model, World Model, and therapeutic journey together, it can do more than answer — it can help them move.

This is what we mean by Psychological Semantic Vector Space: a substrate where meaning is preserved across conversations, where every Avatar inherits a continuous understanding of the person, and where the system grows alongside the user instead of resetting.

“For too long, AI has been built to react to a single message. Persistent Self and AI Mind 2.0 are our commitment to building AI that learns who you are over time — with the rigor a therapeutic system demands and the humility the inner life deserves.”
— Chris Cheng, Founder and CEO of Hamo AI

---

Hamo AI — making minds aware, and awake.

---