𝗪𝗵𝘆 𝗜 𝗮𝗺 𝗕𝘂𝗶𝗹𝗱𝗶𝗻𝗴 𝗱𝗼𝗹𝗹
High-performance AI is easy to access right now. But access is not guaranteed.
Prices change. Usage limits grow. Providers shut down services or change policies. You might lose your AI environment even if it worked yesterday.
The problem is not about trust. The problem is control. Your AI continuity depends on things you do not own.
I started building doll. It is an open-source personal AI continuity system.
doll is not a new model. It does not replace tools like Ollama or LM Studio. Its goal is to protect your personal AI environment. You need your data to stay usable even when models or apps change.
Most people think the model is the center of AI. It is not. The model is just a reasoning engine. You can replace it.
The real core is your state. This includes:
- Long-term memory and preferences.
- Conversation history and sources.
- Research records and project history.
- Permissions and policies.
- Backup and migration records.
A new model might be better. A local model might be necessary for privacy. When you switch models, you should not lose your history.
Local execution is good for privacy and cost. But local apps can cause lock-in too. If an app stops working or changes its format, you lose your data.
doll does not rely on any single format. It does not use ChatGPT or Ollama formats as the standard. Instead, it maps data into a documented representation. If data is lost during an import, doll will tell you.
The rule is simple: Local-complete, cloud-optional.
Your system must work without API keys or an internet connection. Cloud models are just optional tools for better performance. The cloud should never be the source of truth for your identity, files, or memory.
doll is currently in pre-alpha. It is not a daily assistant yet. I am building the boundaries before I connect the models.
I am focusing on:
- Separating memory from secrets.
- Preventing secret leaks in logs.
- Distinguishing facts from inferences.
- Recording where information comes from.
- Requiring user confirmation for high-risk tasks.
I want to define what must survive when technology changes.
If you are a developer, I need your criticism more than your support.
- उपयोगकर्ता की अवस्था कहाँ समाप्त होती है और मॉडल की अवस्था कहाँ से शुरू होती है?
- AI टूल्स के बीच जाने पर कौन सी जानकारी खो जाती है?
- यदि मुख्य मशीन गायब हो जाए, तो क्या कोई सिस्टम वास्तव में खुद को पुनः प्राप्त कर सकता है?
- क्या doll लॉक-इन का एक नया रूप बन सकता है?
प्रोजेक्ट वेबसाइट: https://doll.badjoke-lab.com/
सोर्स कोड: https://github.com/badjoke-lab/doll
पूरा लेख: https://dev.to/badjoke-lab/why-im-building-doll-a-personal-ai-continuity-system-1a1c
वैकल्पिक लर्निंग कम्युनिटी: https://t.me/GyaanSetuAi