long-term memory

Feature Request: Persistent User Memory with Identity & Belief Preferences

📌 Is your feature request related to a problem? Please describe.

Yes. Every time I start a new conversation with DeepSeek, it feels like talking to a complete stranger who has amnesia.
I have to reintroduce myself from scratch:

· My name (Bardia).
· My background (e ‌g., I’m 15, Muslim, and have specific personal or cultural values).
· My deep concerns (e.g., I’ve spent years worried about my health, and I have to re-explain everything each session).
· My ethical boundaries (e.g., as a Muslim, I do not want the assistant to suggest or imply anything against Islamic teachings—like promoting Zina, LGBTQ+ relationships, or content disrespectful to religious values).

This makes long-term interaction emotionally exhausting, repetitive, and far less personal than it could be .

---

🛠 Describe the solution you'd like

I want DeepSeek to have a secure, user-controlled persistent memory system, similar to ChatGPT’s memory feature, with two core capabilities:

1️⃣ Identity & Personal Context Memory

Users should be able to explicitly tell DeepSeek:

· “Remember my name is Bardia.”
· “Remember I am 15 years old.”
· “Remember I am Muslim and want responses aligned with Islamic values.”
· “Remember my profession, interests, or frequent project contexts.”

These memories should persist across all sessions until the user edits or deletes them .

2️⃣ Ethical & Religious Boundary Memory

This is crucial and currently missing.
As a Muslim user, I expect DeepSeek to:

· ✅ Avoid any content promoting Zina (premarital/extramarital relationships), pornography, or sexual immorality.
· ✅ Avoid normalizing or encouraging LGBTQ+ relationships (as this contradicts my religious beliefs).
· ✅ Respect that I do not wish to receive responses that mock or undermine Islamic teachings.
· ✅ Allow me to set these as permanent ethical filters, not something I must repeat in every chat.

This is not about censorship — it’s about cultural and religious respect. Just as some users want a “professional tone” or “casual style,” I want my assistant to respect my faith-based boundaries without me having to re-explain it 100 times .

---

🔐 Privacy & Control

I also request:

· A dedicated “Memory / Profile” settings page where I can:
· View all stored information.
· Edit or delete any memory item at any time.
· Turn persistent memory on/off globally .
· Encrypted storage and GDPR-compliant data handling .
· No automatic learning — only memory items I explicitly ask the assistant to remember.

---

❌ Describe alternatives you've considered

· Manually repeating my name, age, and religious boundaries at the start of every chat.
→ This is tedious, forgettable, and often fails — the assistant still drifts into inappropriate content later.
· Using custom prompts or system instructions.
→ These don’t persist across sessions and are ignored when I switch devices or clear history .
· Switching to other AI models with memory.
→ But I love DeepSeek’s intelligence and free access — I just want it to truly know me.

---

🌍 Why this matters

This isn’t just a “nice to have.”
For users like me — young, religious, with specific ethical boundaries, and a long-term need for emotional and informational support — this feature would transform DeepSeek from a generic chatbot into a true personal companion who remembers who I am and what I stand for .

I want to come back after weeks and hear:

“Welcome back, Bardia. I remember you’re 15, Muslim, and prefer responses aligned with Islamic values. How can I help you today?”

Not: “I’m sorry, I don’t know anything about you.”

---

📎 Additional Context

Similar requests have been made by many users:

· User Profile + Secure Personal Data Storage #912
· Personalization Option (Name, Tone, Background) #850
· Memory System like ChatGPT #319
· Persistent Memory for Users #653

Please prioritize this. It will make DeepSeek the most culturally aware and user-respecting AI on the market.

Thank you for reading, and thank you for building such a powerful model. I believe DeepSeek can lead not just in intelligence, but in emotional intelligence and ethical personalization. 💙