There have been many AI projects recently, take a look at them one by one. I have looked at 0g before and have a general understanding of its framework. Today we are looking at what Sentient @ SentientAGI is. You will find that the founders are different, and even projects with similar directions will have very different focuses. Sentient Protocol is a decentralized protocol that allows communities to build, manage, and use Loyal AI (translated as loyal AI), which should be the alignment of AI, but its alignment is an AI that is consistent with the community, owned and controlled by the community. How was it achieved? Two parts: -The artificial intelligence part, AI pipeline (data management, loyalty training), is the foundation for developing and training loyalty AI. -Blockchain section, governance: a system controlled by DAOs, ownership: tokenized representation of model/application ownership, DeFi: decentralized financial instruments. How can the artificial intelligence part align with the community? He used blockchain to coordinate data management Data management: Through data filtering, instruction data or community determined rewards are carefully filtered and designed to meet community preferences. Data mixing ensures that the model is resistant to manipulation. The data management process is rooted in the community. Enable the community to control the data production of the model through governance models. Users and application builders can submit proposals for model adjustments or improvements, while model owners supervise governance through decentralized voting. Allow the community to contribute, validate, and control training data. Loyalty training: Given the carefully selected dataset, training AI to be loyal and aligned with community values, the loyalty training process includes: Robust Alignment: Train model alignment using supervised fine-tuning (SFT) and reinforcement learning (RL). Fingerprint training: Advanced fingerprint recognition technology is embedded in the model, and fingerprints serve as proof of model ownership, enabling the community to track model modifications and usage while preventing unauthorized modifications. Control training: Implant special queries to maintain model control. These queries ensure deterministic responses to predefined inputs, thereby achieving precise control over AI output. Looking at this picture can better understand his entire structure: The focus of the other teams is on a profitable and loyal (OML) model service paradigm, differentiated from the entire model of closed source models (openai) and open weight models (deepseek). The details of OML will be introduced later. Compared to @ 0g.labs in building decentralized technological infrastructure (including decentralized computing power for training models, etc.), Sentient focuses more on aligning and training models with community will, annotating ownership (fingerprint technology), and a set of innovative models that combine business paradigms (OML model).